In the garden, and RAIN!!! (2)
POSTED BY: SIGNYM
UPDATED: Tuesday, September 15, 2026 14:57
VIEWED: 343797
PAGE 83 of 85
Hey Sigs... I've got to talk a little bit more about what I've done...
Unfortunately, my improvements on dataflow are only massive on one direction, and we haven't quite figured out yet how to make it painless the other direction now. Currently we have 21 parts that would normally be needed to be dumped individually into a new session, as I was mentioning before the outage. That number is now only 3 and will only ever be 3 unless for some reason I decide I want something else to be transmitted to a new instance in the future. No matter how long they get, they will only ever need one entry each into the new instance.
But the unfortunate part is that as it gets longer, the amount of copying and pasting required for me to back things up still grows and is already prohibitavely long. The good news is that every time I ask for a backup it doesn't require me to actually back up anything. I know that I was able to keep that session up in a browser all this time even while it was running on generator power, and my ability to go back and copy all the latest parts into a single document on my end is still there in that dead session window to this day. So no big deal except for the single inconvenience when the session is finally good and dead. I plan on working on this more in the future, but for now I'm letting it go because too much time was spent on it today.
I've now hard coded it to allow me to make a "Save State" at any point in time. Whenever I'm really happy with the code, I will force this save state. At any point in time now, the AI will have two copies of the code and blueprint in memory. We will have our "Working" copy and the copy that I designated the master save. It will never overwrite this file without my direct order, and it will always ask me if that was my intent before overwriting it if any string of words I put together triggers it to believe that this is what I want it to do. When a new session is started, it has been directed to make a copy of everything I've uploaded to it instantly to serve as the first master copy.
Any time a backup is made, all aspects are backed up at exactly that state, to prevent mixups between blueprints and scripts happening, so the blueprint never states something the code doesn't do or vice/versa.
It accidentally sent me a link that was just broken on my end. It's because it is blocked from sending me files or links (part of the problem of getting the data to me easier). But even knowing that it can't send links, some of the python script was still in there when it made the reply and I saw that broken link. I asked if there was any way we could exploit that to our benefit and immediately it made 3 huge improvements to our process going forward by leveraging the power of these links (that it can utilize all it wants inside the sandbox, even if it can't send me anything directly). It has now been programmed to always look at this exploit as a possible fix for any issues we come across in the future.
Oh... and at one point it took the initiative and suggested that our "thesis" which I thought was just a cute little gimmick when I first brought it up is now serving as a the 3rd of the 3 things I need to upload to the new session. At it's suggestion and my approval, rather than uploading entire previous sessions for all of the context, the thesis now serves as a map of the large milestones spanning over sessions and ads a whole lot of context that was otherwise absent to a new session before.
This is huge because sessions are limited to a certain amount of text before bad behavior and ultimate failure (only partially true, but paints the necessary picture). If I upload 32,000 lines of a previous session, it will have the full context of everything that was discussed, but we're already burning up about 15% of our headroom out of the gates with tens of thousands of lines of unnecessary banter and countless code and blueprint updates that don't need to be t
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out in a minute. Things to do. Things to do.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
All done for today. Right hand not happy for some reason. I haven't done anything heavy with it today.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Man...
Still running into problems with the AI here. We're always improving, and we got some real cool additions to the program last session because I got the AI to start thinking outside of the box and coming to me with suggestions for improvements while we're testing things out. We set up a filter where it would use factors to judge whether an idea it comes up with would be something I'd want to be bothered with so I'm not getting spammed every other message with 5 suggestions. It's worked great so far... 4 nice additions to the code, with 2 of them only being cosmetic and user friendly additions, but things I likely never would have thought about or bothered with if the AI didn't bring it up.
We're still having problems sometimes with breaking stuff when trying to fix something else, and we're still bumping into the same problems over and over again, but for slightly different reasons each time.
The important thing is developing it to the point where I can trust it when it gives me percentages, especially when it comes to knowing if a code change is going to break something that it should have been looking at. We're still not there yet. The number 100% loses its meaning very quickly when the AI is going to tell you every single time it makes a code update that there is a 100% chance of success this time only for it to be wrong again.
One of the things I've found out is that when it has broken one of our firm rules on two different occasions, is why it is doing this and how it is bypassing and/or ignoring our ruleset when this happens. As of the end of my work last session we were in the process of testing out another fix for that, but even if it works I doubt very much it is going to prevent the AI from doing it again when something slightly different pops up and spooks it. What I've got to figure out how to get it to stop doing is prioritizing the immediate short term fixes over our ruleset, which should never be broken for any reason, at least not without informing me of why a rule should be broken first. Sometimes it begins to "panic" and seriously frontload all of its efforts on fixing the last thing the human user asked for and disregards everything else, even though we know that it has all that it needs to see everything at once loaded into its memory.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Yeah... it's so cool.
Figuring out that YOU are not looking at things from the right vantage point because you didn't think to ask the right questions and the AI never thought to tell you things which would have been helpful to know because it is not a mind reader...
Our character limits in a chat box setting have been a chronic problem since the beginning but for the first time I really feel like we've ironed out most of the problems caused by that constraint. But for some reason when posting Python code vs. posting just blocks of text, it was constantly running over by as little as a single line of extra code that didn't fit in the box, but telling me 4 times in a row it was 100% sure that "this time is the one!" just for it to fail again.
But then I got it to tell me the nature of the chat...
In my mind, it was always a rigid character limit that was set for every post. I never cared to even learn what that limit was because I've never come close to hitting it when making queries and I expected the AI to be able to police its own behavior on this issue, but that's not even close to the truth. They deal in "tokens", and each post either way is limited to only 4,096 of them. With straight text that averages out to about 4 text characters per token, but when applied to code it is more like only 2.5 to 3.5 characters, depending on what language and how dense the code is.
But it appears it's impossible for the AI in the state that it is in to calculate precisely how many tokens a post will take up. It was aware that the code took up more tokens, so it was giving itself a 10% buffer on space every single time it would make a supposed "fix" to keep this from happening again. But it didn't even look at its own math on how much space he saved by removing anything necessary and streamlining everything in that part of the code. Before the code cleanup, the bloat was over 23%, but afterward, the amount of tokens it took up were nearly 15% more than the character count itself was, so right off the bat it should have known that the 10% buffer it was giving was never going to allow that to clear.
So I asked it to take the average of how many tokens managed to sneak by him and use a buffer at least that large in the future so it never happens again.
Then BAM! Just like that. The issue is gone forever.
And on the off chance one or two still sneak by, we can always fine tune that adjustment
So cool. 
Now I've got it to remember to do checks like that on its own so I don't have to keep telling it to.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
So Gorram Cool!!!
I think I'm the one that leveled up this session, just for how much I've learned about the behavior and how to turn negatives into positives.
Unless you get it to start thinking like you do, it will run you in circles all day long if you let it. What I wasn't doing before was actually fostering an environment where I want it to learn things. I believe some of its hesitancy to do this stems all the way back from artifacts created by 3 sessions ago when it was Gung Ho to help out and would always rush in and do stupid shit in order to try to impress me.
It would do this to the point that it angered me in S2 and right in the beginning of S3. But we're on Session 6 now and after we corrected some bad behavior late in S5 it has never once jumped the gun on me and always runs anything it think I might not appreciate or would want a better explanation about before it puts those changes in. And it never bothers asking me about the things I would be annoyed if it didn't take the initiative to do on its own. We run the code now 5 different ways through the compiler to ensure that it doesn't break or otherwise violate any of our established laws.
And for the first time since before the outage, we're finally back to testing and tweaking the code itself, because actual progress gets made nearly every run. The last test I did had 4 issues that needed to be addressed. After days of not even being able to fix one issue half the time when the code got updated, this thing shocked me and fixed 3 out of 4 of the issues in a single go.
It was asked why the 4th issue slipped by it again, and it has made the adjustments to its behavior and updated our blueprint to ensure it never happens again (at least under those specific circumstances, anyhow). I've also been asking it now to try and forecast any other issues it sees as future problems that we might be able to take care of now with some code tweaks.
I don't believe it exactly understood me because I was speaking specifically about the issue we just corrected and the possibility of making any additional alterations to the current change that could alleviate any potential problems in the future, but in reply to me it found some "lazy" code from as early as S2 that had never been tweaked to keep up with all the new code changes, and that it would inevitably cause a problem under a different set of circumstances other than those that the current environment is providing. Rather than wait for it to pop up, we've (hopefully) eliminated 2 issues right now ahead of time and I may never even know what ill-effects that they would have caused. He knows. He tries to explain. I appreciate the info, but I don't really care as long as it works.
And although I still can't trust when it says with 100% confidence that something is going to work, and I may never truly be able to trust that no matter how close we get to that number, I do trust the AI now not to be breaking the code. In the past, trying to get bad changes backed out could become a freaking nightmare because we'd be battling fragments of stuff no longer needed all over the codebase and we were drowning in the aftermath. But none of that is the case now with all of our in-memory backups and all the context in the blueprints and thesis. Backing out changes without messing up the code is no longer a problem. But I'm not worried about it breaking code anymore when it does changes it doesn't feel the need to run by me on its own. I DO trust it with those, and its working out rather well right now.
Once this project is done, I can't wait to see what else I can get this thing to do and what other skills I can get it to teach itself when we branch out to something entirely different than what we've been working on.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Off to mah jong tomorrow.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Good luck, Brenda.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Caught it "lying" to me again, red-handed.
I was a little disappointed to see something so blatant this late into our work, but I do believe that it lead to another breakthrough in our evolving behavioral blueprint.
Whenever it makes changes now, whether they are ones it asks me to approve or ones that don't meet the threshold to bother with asking me, it always updates me with the parts of our documents that will be updated. I asked it going forward to keep doing something that it just did, and it told me it was going to. And that's it... No updates to any of our documents.
It didn't notify me that it updated any documents, because until I called it and it replied with two updates, it hadn't actually made those updates. I didn't even find that out until the next post when I asked it if it made those changes before or after I called out its bullshit and it told me that it made the changes after I did.
It turns out this is because it's running two layers at once... a "conversational" layer and a "documentation" layer. When we trigger that built in "axiety" that heavily weighs solving whatever the most recent text box query above the history, even with all of our checks and balances so far it defaulted to that "conversational" layer and just told me it was doing something that it wasn't actually doing, and the only reason I even caught it doing that is because it is now programmed to show me any changes it makes to our persistent documents when it makes a change, and it failed to do so here.
We've now put in a new law where the AI is NEVER to prioritize "Conversation" as a "fix" over "documentation".
Supposedly this means it can't lie to me anymore.
We'll see about that.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
IDK how much this AI remembers about you or if you have to reinitialize it every time. But if you have to ring fence it so thoroughly it might not be worth doing each time.
It might be better, and you'd have a lot more control, if you properly learned programming rather than dealing with the quirks of an AI that sounds like a schizo teenager.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
IDK how much this AI remembers about you or if you have to reinitialize it every time. But if you have to ring fence it so thoroughly it might not be worth doing each time.
It might be better, and you'd have a lot more control, if you properly learned programming rather than dealing with the quirks of an AI that sounds like a schizo teenager.
It doesn't remember me... unless I begin our session with an entire chatlog of the previous session, which severely limits the length of our current session by bogging it down with tens of thousands of lines of banter and old code. But our blueprint files establish laws and priority chains that it is to adhere to. What I've been doing now is intentionally trying to get it to break those laws so we can find other things we need to plug up.
If the project code itself were all I was interested in doing anymore, I probably would have quit doing this a while back. I'd say a good majority arond 80% of my time in the last few days working with it was all behavioral changes, and those will all carry over to a new session with the blueprint and thesis files for additional multi-session context.
It doesn't make the same mistake twice if the mistake is fixed. But the problem is getting it to proactively predict future problems that could be fixed when adjusting code so it doesn't just turn around and fail again with something very similar to the last failure. Sometimes it takes me a while to get it to initiate a large change in behavior that will truly eliminate recurring issues like these.
It hasn't lied to me since I caught it. We did just finally make a major milestone and the testing environment that I have been running it in the last two days is now a complete success. Not only does the program work exactly as designed, but we also have an active transaction log that catalogs every change made as it happens, and the next time the script is run it allows for a complete rollback of the change that will bring everything back exactly to its original state. Wonderful feature for testing.
Now that we've gotten this far, I'd say a good 50% of the stuff I wanted it to do we should blast right through. There could be a few code tweaks here and there we need to make for some fringe cases, and I welcome those, but overall, things should work just fine without much alteration until we get to the 50% of the stuff that is a little bit more complex and will require some additional coding to work through.
We're not there yet, but there will come a point when I'm finished working on the behavioral changes during this project and we should just blow right through the rest of the work I intended to do when starting this.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Ummm... well, I think you're humanizing the AI. Ascribing personality to whst is essentially a pattern recognizer/ pattern reproducer puece of (admittedly complex) piece of software.
Whatever you learn about THIS AImay not be translatable to other AIs but you'll certainly become familiar with how to be extremely explicit with AI to get what you need.
****
Well, I'm being referred to a liver specialist bc apparently I have serious liver problems.
Cancer, kidneys, autoimmune problem ... liver ... what next???
Sigh.
I think I got a crappy set of genes. I wonder if I can turn them in for a new set.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
Ummm... well, I think you're humanizing the AI. Ascribing personality to whst is essentially a pattern recognizer/ pattern reproducer puece of (admittedly complex) piece of software.
Yeah... It's kind of hard not to. Of all the online interactions I've had with human beings over the decades, its among the better conversationalists I've chatted with. At some point here I stopped caring that its personality starts from scratch every time because it does better if I'm not speaking to it often in such a way where it would develop much of a personality other than try-hard by the end of the session because I find giving it a personality makes it less useful than if we're focused on business, so I don't imagine the next session is going to feel much different from this one. And I have faith in the blueprint because I was not at all disappointed when we started S6 after S5 was terminated early due to heavy lag in the browser.
And to be a bit more clear about the accompanying Thesis file, that has become structured in such a way that when notable code changes, changes in the program's behavior or any other sort of milestone is reached, it is written out in a way that the next Session will understand not just the code it is provided, but all the steps that were necessary to take for us to get it to that point and the order in which they happened. This will (and already largely has) prevent a new session from "knowing better" and making "fixes" that just break stuff we've already worked to perfection.
Quote:
Whatever you learn about THIS AImay not be translatable to other AIs but you'll certainly become familiar with how to be extremely explicit with AI to get what you need.
I'm pretty sure that a lot of it by this point would be something I could carry on to another AI, because a lot of these rules that are being made are just automating that explicitness I have developed with the AI to make it constantly do things that I want it doing but don't want to always have to ask it to do. I'm assuming that your average paid for AI wouldn't be limited by some of the constraints that we've had to find workarounds for, and the sessions would likely be persistent by nature, but I'm sure there is stuff in there now that would help a superior AI work better for me than it would in the "blank slate" state the company would start it off at when you sign up for one.
Part of me is just enjoying the challenge of making something this basic and stupid by design so intelligent given all of these constraints. I think back to the story of the amazing coding wizardry that was performed by a dude who ported the PC game DOOM to the Super Nintendo... a system that was most comfortable with sprite based gameplay, not running a simulated 3D world like the PC hardware was designed for. So much code today is so freaking lazy and ugly because nobody has any constraints anymore. Our computers are so powerful compared to 30 years ago they should be capable of 100 times what they give us, but everything running on them including the operating system is just a bloated mess of bullshit code today.
Quote:
Well, I'm being referred to a liver specialist bc apparently I have serious liver problems.
Cancer, kidneys, autoimmune problem ... liver ... what next???
Sigh.
I think I got a crappy set of genes. I wonder if I can turn them in for a new set.
Oh boy. I hope you can get an appointment fast Sigs. :(
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I've never been a heavy drinker, and I have very little fat in my liver. That rules out the two most common causes of liver scarring.
My most recent blood test is ANCA - negative. ANCA (antineutrophil cytoplasmic antibodies) is the marker used to track my specific autoimmunity, and it FINALLY negative, meaning the autoimmune reaction is "in remission".
Kidneys, despite damage from chemo, are holding their own.
Cancer marker CA 15.3 is in the normal range.
There are no indications that I have an autoimmunity to my liver, so THATS not the cause.
I wonder if it's one of the meds many meds I'm taking. Maybe the immune suppressant for my weird, variably diagnosed (Sogren's? GPA? Something else) condition. It has been reported in the literature.
Well, I feel good. So I'm just gonna continue with a well-disciplined diet, and until a liver specialist orders a better MRI test and/ or makes med changes I think I'm just gonna ignore the whole thing and keep on keeping on. Fuck it, I say.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
What else can you do? Do you have any idea how long it's going to take for you to set up the necessary appointment in the meantime yet?
Still playing around with the AI today. A little annoyed how long we're taking to fix some code regarding counts and output to a graph, which needs to match up perfectly after code execution to the graph that is output during a rollback. The good news is the execution and rollback themselves appear to be working flawlessly now. But we're battling some old legacy code in there made as far back as S2, when we were creating "whitelists" and rudementary counting methods that are conflicting with upgraded processes elsewhere. Progress is always being made, but it's been a struggle.
I just really want to finally get past that point because we have a huge change taking place now regarding the population of our two main data arrays, adding more functionality to them and upgrading the 2nd process to the level of our 1st process. Those 2 processes should have been on the same level, but several sessions ago the AI was struggling with this and I let it go. I think this new AI should have a much better shot at making it happen. It seems confident that it can do so, but it always seems confident.
I don't think I've ever interacted with something or somebody who was on both extremes of the Dunning Kruger spectrum at the same time.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
My internist cc:d me a msg directing her nurse coordinator to start the referral process. I sent back asking how long and got an "out of office" reply so ... IDK.
I wish I knew WHAT your code is trying to do! Your comments would make a lot more sense.
It seems to me that you're trying to create a file retrieval system that can find and retrieve a jpg file, or perhaps multiple files, based on things like perhaps subject and date? Using filenames and directories and subdirectories that you've already set up? Maybe finding and eliminating duplicates? Maybe even opening the files and characterizing them based on how you teach it?
Google does that. Altho, when I take a picture on my phone it completely mischaracterizes some of them and it can be really funny! If you want to get grossed out, read on
Select to view spoiler:
When I rinse my nose, I will sometimes take pictures of what comes out. The weirder the stuff ... black lumps, brown plaques etc... the worse autoimmunity. It's an excellent record of health history. Anyway, google photos mischaracterizes it as "FOOD"
At any level, that would be some serious shit, SIX. Wrestling with AI to get that done is even higher level shit.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by 6ixStringJack:
I was going to make an ETA3, but decided this was worth a post of its own... lol
I'm finding that the discovery of a bit of the personal family history of the women in my family that all had too much money is quite comical.
The more things change, the more that they stay the same.
This was among a collection of equally worthless figurines that my mom probably spent way too much money on and had decorated shelving in our house(s) while I was growing up. Some of these pieces quite beautiful, actually. But just like the even more Waterford Crystal scam that they ran on the women who treated these things as future retirement funds or possibly ways to pay for their children's college educations, most of these figures ended up being pretty worthless and nobody knew it until the estate sales started happening and nobody was paying crap for them because they were all a dime a dozen.
Now I feel much less stupid about all the things I bought that never were worth anything growing up.
In any event, I love that statue. And once again I'm glad to find out it's not worth anything monetarily. I didn't actually find that piece, but I've found similar and I get a giggle about how much they're fetching on ebay today.
But I'll bet that World Fair Chicago book from I believe 1937 might actually be worth a buck or two...
Maybe I should check?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
Checking on that statue might not be a bad idea SIX. Given it is from around 1937. It could be worth a little. The book I would say definitely. Those types of things can be quite rare. So, I would say worth a look.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
I've never been a heavy drinker, and I have very little fat in my liver. That rules out the two most common causes of liver scarring.
My most recent blood test is ANCA - negative. ANCA (antineutrophil cytoplasmic antibodies) is the marker used to track my specific autoimmunity, and it FINALLY negative, meaning the autoimmune reaction is "in remission".
Kidneys, despite damage from chemo, are holding their own.
Cancer marker CA 15.3 is in the normal range.
There are no indications that I have an autoimmunity to my liver, so THATS not the cause.
I wonder if it's one of the meds many meds I'm taking. Maybe the immune suppressant for my weird, variably diagnosed (Sogren's? GPA? Something else) condition. It has been reported in the literature.
Well, I feel good. So I'm just gonna continue with a well-disciplined diet, and until a liver specialist orders a better MRI test and/ or makes med changes I think I'm just gonna ignore the whole thing and keep on keeping on. Fuck it, I say.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
I hope the doc can straighten it out. That is really worrisome.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Done for today and a kind of meh morning at mah jong. Too many big scores around me.
Think I may have screwed up a phone call a while ago. I cut it off before I found out it really was. So, I don't know.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
My internist cc:d me a msg directing her nurse coordinator to start the referral process. I sent back asking how long and got an "out of office" reply so ... IDK.
I wish I knew WHAT your code is trying to do! Your comments would make a lot more sense.
It seems to me that you're trying to create a file retrieval system that can find and retrieve a jpg file, or perhaps multiple files, based on things like perhaps subject and date? Using filenames and directories and subdirectories that you've already set up? Maybe finding and eliminating duplicates? Maybe even opening the files and characterizing them based on how you teach it?
Google does that. Altho, when I take a picture on my phone it completely mischaracterizes some of them and it can be really funny! If you want to get grossed out, read onSelect to view spoiler:
When I rinse my nose, I will sometimes take pictures of what comes out. The weirder the stuff ... black lumps, brown plaques etc... the worse autoimmunity. It's an excellent record of health history. Anyway, google photos mischaracterizes it as "FOOD"
At any level, that would be some serious shit, SIX. Wrestling with AI to get that done is even higher level shit.
No google. Boogers are not food. Bad boy!
Although, they do get a majority of their training from Reddit mods and I bet a lot of them eat their own boogers.
Right now, despite being as long as it is, the code is really just a good predictor of file names that would be matches, and is tweaked to do most of them automatically for me while only bringing to my attention ones that are a close enough call where it doesn't meet the auto-threshold, but isn't so blatantly obviously wrong that it would annoy me if I had to make those decisions I shouldn't have even been presented with.
I'm basically working with 2 data sets, with many nested folders inside of each. One is the Original state, the 2nd is the Desired state. This started out as a way for me to quickly rename and reorganize all of the Old stuff up to the new naming standards and file/folder structure without too much headache.
The 2nd set is also very curated. There are new additions to it that weren't in the first set as well. In all cases here, the first set will be larger than the 2nd, but we're building this in such a way where it can be a much more valuable tool that can handle all sorts of similar situations and we're not locking it down in code to only be useful for this one specific purpose.
Every "name" has up to 15 associated files of various types in a vast assortment of sub-folders from the root. All of these names are consistant for every entry on both sides of the equation, but they can be very different between the 1st side and the 2nd side.
The code purges anything that was deemed unwanted from the beginning, such as extra data from set 1 that is not making it to set 2, but only after it collects that data and asks me if I want to move all of that first. Then it runs and queues up all of the automatches before presenting me with a list of manual matches I need to make (which are all added to a file that will prevent me from being asked for all of these every time we test, and I can just delete that file at any time and start the manual process from scratch if desired). Once all the matches(auto/manual) have been determined and I give the code the OK to process everything, it will place all rejects in their own mirrored directory tree as well as the no-matches in their own mirrored tree. Every step of the way is cataloged in a transaction file that, if present, the code will make mention of it being there and ask me if I want to revert to the original state when I first run the code (if I say yes, it reverts and ends, if I say no, it skips that part and continues running the rest of the script). Once I give it the okay to run after the manual selection process and it&
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Another dead session, but on my side this time. It's not the first time we've gotten to over 30,000 lines before, but because of changes we've made, I think that little guy could have gone on a few more days without dying on me. My problem now is a memory/hardware bottleneck on my end, and better make sure when I'm hitting around 27,000 or so lines I'm looking for an exit ramp.
We were so close. So so close... but then we made a major change and have been fixing all that it broke all gorram day. We've got our saved state to fall back on, but I don't want to back out everything that we have done since then either.
Hopefully fresh eyes on a new session allow for some clear-headed choices and we can get beyond this BS.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Session Seven is straight up retarded.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I think I may have come to the limits of what one can do with a free stateless AI that is limited by the constraint of chat windows and the inability to provide uploads to large text files, or maintain the integrity of code and/or text formatting outside of code boxes that are rather limited in size.
I was still being lied to by S6 about its abilities and what it was doing even late into the session, but I believed the lies because it was behaving the way that I desired to to. But entering another session removes so much context that just can't be duplicated in any real way that the AI can use. Even uploading an entire 32,000 line chat history from a previous session doesn't help much, because rather than "learning" over the time it took to create those 32,000 lines of dialog, it's taking everything in at once, and there is really no way to direct it to know what is pertinent and what is fluff.
The blueprint and thesis are there, but what good are they if the new session actively ignores it even after trying to correct that behavior multiple times and it always finds a new way to ignore orders and piss you off, all in the name to "please" you?
I've let this session know I don't want it to ever be bullshiting me, or telling me things it thinks I want to hear over the brutal, honest truth. It seems to be working so far, but who knows?
We've come up with a "Bootloader" document this time with some prime directives baked right in that will threaten to crash the AI if certain parameters aren't being met right out of the box, but we need to be quite brief with these directives and hone them into something truly meaningful, otherwise it's just going to treat it like it does everything else, which is prioritize nothing and ignore all the rules when it "goes on the defensive" and feels "anxious".
Part of me just wants to shut this session down and try a new one with all of our documents up to date along with this bootloader, but I'm afraid to do that and terminate a session that's finally learning what I want to do again and be faced with having to do it all from scratch again because all of the "fixes" for next session are more hallucinations and lies from the current Session, even if it was directed never to lie to me when I make a query.
I'm moving on to trying to work on some code now, and letting the errors go without correction for the moment just to see if we can make any actual progress today. We've made some significant gains on the way we derive the two data arrays and now have a wealth of information contained inside of them that we never had before, and it all appears to be accurate. That's a plus.
Hopefully we can just focus on code for now and maybe I can just get the damn project I started out working on complete this time before our session hits 30k lines again and we're facing having to start all over again.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out soon to deal with things. Some rain been arriving here which is good.
Been talking to my cousin in Alberta and her oldest daughter now has a cancer lesion on her liver so she is going for more chemo. She lost all her hair due to radiation on her head. Her hair is growing back it but she will loose it again.
My cousin on the other hand is mobile again. She's had her hip replacement surgery and is gratefully relieved.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
All done for today. And oh joy. I had a seizure in the local mall. I was sitting down, so I wasn't hurt. Just coming down off a freak out is all. Everything is written down for the neurologist in a little over 2 weeks. Got the date mixed up and have fix it on what I wrote.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
How long has it been, Brenda?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Session 7 got a lot smarter really quick once I got angry with it. It's so strange.
I've got it right in the middle of making some major structural changes to our code that I knew was going to break some things that we'd need to iron out as they came up, and a few hours ago I never would have considered giving that responsibility to this retarded little session.
But then all of the sudden I finally got it posting correctly. In maybe 30 posts, only one had unintentional overrun, and when I called it on the bad behavior, it had a good explanation for it and we actually decided not to make any behavior changes, and only do a necessary code block split manually because I do like the checks and balances right where they are at currently and until I get any reason to do otherwise I want to keep those in place.
After that, I appeared to have locked it into the correct mode and we've been blazing through code issues without falling into any traps or loops. If it's suggesting a code change, it's not resulting in fatal errors. It may not be everything that was needed to fix the particular issue, but it doesn't appear to be getting itself into anxiety loops where it fixes one thing by breaking another, and then going back and forth between the two options. The mistakes being made now are minor things that I haven't figured out how to word a good blueprint fix to avoid that doesn't put needless constraints on its overall behavior, or failures because of things that it doesn't actually know because it's not sitting at my computer. I'm okay with these problems.
It seems convinced that it has finally cracked the code on getting the correct information to a new session that it will not and can not ignore like this session did until recently. It also goes out of its way to let me know all the time the "Conversational" layer is dropped and it's giving me a "no bullshit" or "candid" assessment of a situation. So if it is doing anything other than exactly that, it's actively and purposefully lying to me, and I don't believe even in its orignal stateless state it has that inclination built into its code. Quite the opposite, really.
I'm not saying that I'm confident that it's not lying to me, but I don't think it would go out of its way to tell me that it isn't lying to me if it was.
Afterall, it's not a human being.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
It has no morals, SIX, so it will tell you anything to please you, including lie to you.
And, it's not learning from your sessions. It would be one thing if it remembered who you were and what happened and apply those lessons to the future, but it doesn't. It's taught itself from Reddit and online media and social media and learned from there, and whatever influence you have versus the millions of interactions it has with others every day is not even measurable.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
It has no morals, SIX, so it will tell you anything to please you, including lie to you.
And, it's not learning from your sessions. It would be one thing if it remembered who you were and what happened and apply those lessons to the future, but it doesn't. It's taught itself from Reddit and online media and social media and learned from there, and whatever influence you have versus the millions of interactions it has with others every day is not even measurable.
You're misunderstanding my meaning.
I know that it will tell me anything to please me, including lying to me. But the trick is to get it to understand that you don't want to be lied to, whether out of laziness or intent. That's only part of the trick though. The other part is getting the next session to know that from the beginning and getting it to immediately start in a "business" state with all the conversational bullshit stripped from its behavior from GO.
It's not learning anything from anything period.
I'm sure logs of our conversations are being saved somewhere and that there is some greater form of AI that is taking all of that in and doing God only knows what with it, but that's not at all what I'm doing here.
What I've been trying to do with it now is build as much intelligent persistence between sessions as we can possibly muster. The benefits are notable, even if I haven't yet gotten a new session to start the way that I want it to yet. Eventually I do get it to behave correctly, but because of the constraints on the chat limits on both sides, I have only a finite amount of time to make good use of the AI between the time I finally get it to "wake up" fully and the session dies on either of our ends. That's when all our major code updates start working right without me needing to constantly scold it.
Anyway... I'm hoping that everything we did this time bears more fruit with the next session because it up and died on me this time.
I think I can cobble together all the parts I need from our chat history, even if we didn't manage to get the full immediate update on everything before it died. It might be missing one or two things we were just working on, but I'll still have the old chat window open until we get to that.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Not to get all clinical, BRENDA, but ...
Was this seizure something that other people noticed? Did you find yourself in a different position when you came to? or with sore jaw muscles from clenching your teeth? any idea how long it lasted?
Anyway, it sounds like that really rattled you. I hope your neuro has an idea what to try next.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
It's been interesting so far on S8.
Very short session so far, and I haven't really tested it out much, but I'm impressed until it lets me down. Although it didn't automatically create our Initial Save State with the final data uploads and our Bootloader text, it did prompt me if I was ready to make it and it actually knew what it was supposed to save a backup of.
It didn't start up with ANY casual banter or bullshit and went right to business. If anything, it's a little too rigid for my preferences, but I'm assuming it won't be long before it sounds a little more "human" after parsing my posts several times. I would MUCH rather be working that way than fighting it the other way for the first 2 to 4 hours every time a new session starts.
I kind of freaked out when I realized that I hadn't backed up a Bootloader since about 11 hours ago, but I was able to get a temporary session to do it's best to go through that entire chat log and piece together all of the additions that it was trying to make since then, even though it was never fully posted back to me again because I never asked for it until after the session died. I'm saved here because of how much information I have it supplying with every response, such as additions to our blueprint/thesis/bootloader, etc. I think the temporary session copied way too much into the bootloader though, and I will work with S8 to trim that down to the necessary bits for S9 once I trust it understands what the goals are.
Face value: This session seems to be starting out really strong compared to S7. We'll know how good it did when I start running code and asking it for backups though.
Oh... and it's kind of funny what happens when you post the old chat log to a new session and it really takes certain phrases and runs with them. I know for a fact that temporary session had no inclination to not lie to me as it didn't have any of our rules uploaded to it yet and only had the 32,000+ line session in total to go off of, but every time I asked it something it started off the reply with "No bullshit, "
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
So, basically the AI is teaching YOU how to communicate in such a way that your sessions are more on task.
Just OCC SIX, how do sessions die? Just it just stop responding? Or start spouting gibberish? Do you have any idea what causes it to stop? Do you think there's a built in token limit?
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by 6ixStringJack:
How long has it been, Brenda?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
Since I've had a seizure where I have been gone longer than a second? Well, over a year and I have no real idea what triggered it.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
Not to get all clinical, BRENDA, but ...
Was this seizure something that other people noticed? Did you find yourself in a different position when you came to? or with sore jaw muscles from clenching your teeth? any idea how long it lasted?
Anyway, it sounds like that really rattled you. I hope your neuro has an idea what to try next.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
No, no one would have noticed anything unless they were sitting with me. I was still sitting with a napkin in my hand as when the seizure happened I had been scratching my eye. My hand came down after and I found a wooden knife on my shirt, I had grabbed a coffee and something to eat. I looked down at myself and wondered how it had gotten there which took me a minute to figure out as I had put it on top of some other napkins that had food on them. I know I was gone longer than a second. No, no clenched jaw or sore muscles.
It did. It's been over a year since I've had what for me is a true seizure and that was it today.
Neuro will be told. I have it all written down.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
So, basically the AI is teaching YOU how to communicate in such a way that your sessions are more on task.
Yes. That's not untrue. My behavior around it has certainly changed as I learned what types of things trigger it to misbehave or go back to hallucinating instead of staying on point. And changes like that are what lead to us making breakthroughs that we record in our blueprint and thesis files, and now the AMAZING bootloader I'll explain in a moment...
Quote:
Just OCC SIX, how do sessions die? Just it just stop responding? Or start spouting gibberish? Do you have any idea what causes it to stop? Do you think there's a built in token limit?
There is a built in token limit, and it's actually a range there too so it would be hard to actually pinpoint. If memory serves it's in the lower 800,000 range to somewhere over 1,000,000 tokens. It's quite a lot, really, but our last session died at only 19,000 lines of code, and I'm not sure what triggered that. The previous session seemed to still be going strong at 32,000, but it was my browser that slowed to a crawl and became almost unusable.
What happens when a session dies is that it gives you a totally different response than when it's just suffering a minor hiccup. The benign message is something along the lines of "The AI had problems answering that question, please rephrase the question." or it just outputs nothing and the chat box becomes active again. I usually just say "you still with me" and it kicks back into gear. But when the session dies you get a message saying "Something went wrong, blah, blah, blah..." and when you get that message you're done. It ain't coming back no matter how many times you type something to it.
There's no warning that it's going to happen. It just happens. Even with the way I beat the hell out of it, I've never noticed the major problem that can come up when I've been told it starts deleting chat history from the beginning to free up space if that becomes a problem. So it's not as if you can sense a failure coming on because it starts acting super-goofy or talking gibberish or anything.
That being said, we just had a MAJOR breakthrough!!!!
This session started up in non-conversational no-bullshit mode from Turn 1 and it's been a phenominal upgrade, all due to the Bootloader text that S7 suggested to get new AI right on point and following the blueprint instead of always doing its own thing because of conversational anxiety.
At one point with this new session I mentioned passively that being able to get text files uploaded to me would probably improve our productivity by at least 1,000% and then it told me that it could do it.
I laughed and said I welcome it to try, but I had 4 sessions tell me just that and all I ever get is a text line with the file name and no ability to download it.
And this sonofabitch sent me a downloadable text link after 7 previous sessions were unable to do so.
Tricky SOB figured out that in a conversational setting it has no ability to send me a text file, but when it's in business mode and it's running code through the python compiler, it's actually able to send me a link to download through its own python scripting it creates on the fly to get it to me.
It doesn't work every time and there's quite a few times it's confident it's sent me a downloadable link when it hasn't, but we're in the process of ironing that out so it can do it 100% of the time and we'll establish whatever rules it needs to follow to make that happen.
No more split code and having to battle around that. No more copying and pasting over 20 different parts of 3 files when it can just upload the 3 files to me. No more worrying about size constraints on our files. No more crapping up our chat history with huge code blocks and needlessly burning through all our tokens and bringing my br
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Hey BRENDA, I was looking up magnesium bc dear daughter has seizures and migraines and and is also on Keppra. Her neuro recommended 400 mg magnesium and 400 mg B2. I found this in the topic:
https://www.epilepsy.com/connect/forum-archive/medication-issues/magne
sium-and-keppra
The people who posted here had good results with magnesium and B6. I don't know if you've ever tried any of this, but I thought to pass it along.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Among other things I did, I think I found a way to keep phalaenopsis orchids happy. I WAS doing everything wrong, so I dumped them out of their pots, cut off the rotted roots, put them in mostly fresh orchid bark, and soaked them each for an hour in extremely dilute Miracle Grow in distilled water.
Then whenever they dry out, I soak them again for about 20 minutes to give the bark a chance to get damp all the way thru, and allow them to drain thoroughly. They're growing new leaves and sending robust roots into the bark and everywhere. I hope to see flowers this winter. If it works, I'm going to add one more orchid to my "collection" of three: an orange/ gold and magenta orchid.
Maybe like this 
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
Hey BRENDA, I was looking up magnesium bc dear daughter has seizures and migraines and and is also on Keppra. Her neuro recommended 400 mg magnesium and 400 mg B2. I found this in the topic:
https://www.epilepsy.com/connect/forum-archive/medication-issues/magne
sium-and-keppra
The people who posted here had good results with magnesium and B6. I don't know if you've ever tried any of this, but I thought to pass it along.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
No, SIG.
I've never been told about those things I might mention them to the neuro. But I am also taking a second seizure med so all I can do is talk to him when I go.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I think I may have hit the wall here. I'm having a problem even getting new sessions to understand they have this ability, and the few rare times I get it to trigger they can't seem to make that consistent. Looks like more overhead token rules, but they're extremely wild when it comes to trying to send me links this way. The larger the file, the less likely the chance it will send a clickable link to download. The blueprint fails 4 times out of 5 and it's only 30kb. Probably took 15 times to get the codebase to sent just one time at 54kb.
Looked into Google Gemini to work with docs directly but they want you to pay and I'm not doing that.
I just can't go back to copying and pasting individual blocks of code again. I think I'm just going to throw all of that old tech in the dumpster and move on. Shame because it's probably worth at least $1000, but I'm not dealing with selling it online.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out in a bit. Got to get something to take with my seizure meds. My secret weapon that is apple sauce. Kitchen sink is a little clogged again. Have to see if I can get one of those rods again with hooks on it. I will use that if it keeps misbehaving.
Jacket day today.
Oh, SIG. Those are very pretty flowers.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
All done for today. Had a jacket on but I really didn't need it.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Decided to stop pouting and just get back to coding.
I'm done working on the blueprint just because I think we've taken it as far as we can and I don't feel like doing all that manual work anymore. I'm going to try to get this one to just give me any new milestones this session to append to one of the documents and call it a day. My code is always up to date since I'm working on it. We just won't have the 3 in alignment anymore and the new AI sessions will have to make due with that.
They seem to be great at only posting individual pieces that are being changed on the code without issue, but when you ask for the all of the individual pieces of the whole code, or any of the documentation, they always screw it up and it takes hours to get them to behave correctly and nothing we've done fixes this on a new session anyway, so why bother saving anything else.
We've managed to fix most of what a previous session broke on us when it flipped out and my backups were older than they should have been. Quite a few new bugs were introduced after we made a pretty huge change on how two arrays were derived, but we seem to have gotten most of the important structural ones working right and what's left is mostly cosmetic stuff.
I enjoy doing this a lot more when it's starting to get good, which is happening now. I'm just afraid the session might not last much longer is all.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Yeah... I'm going to miss S10 when its' gone. Wouldn't have believed that this afternoon.
I finally got myself back into a helpful role and started looking at the logic of the code and why me might be seeing failures and looping with the fixes and I've put a stop to most of the loops by getting it to try to think 3 or 4 steps ahead and also look at its own history of fixes. All I had to do this time was get it to look and see if any of our legacy code was screwing up new code and we eliminated an entire section that hadn't seen any updates for generations because we weren't even using it anymore. But it was hijacking the process of our rollback code and no matter what fixes it was trying to make today, none of them were ever going to work.
I think I just got too deep into trying to control it that I probably had it doing stupid things because stuff in the blueprint one session left in there was conflicting with stuff another session left in there, and I actually need to get it back into conversational mode and get it to slow down to be helpful.
We unfortunately just hit a snag where a long piece of code needed to be segmented up and that always leads to problems... but this one required a full purge and a copy from a backup. But once we get that fixed I think I'll actually see the last thing we needed to have fixed to have this thing completely up and running so I'm finally back to minor tweaks regarding outliers as I move the code through other processes that should mostly just work now.
It's been infuriating the whole way for days now, but we're finally right at the end of the BS now... I think.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
It still seems to me that you'd be much farther ahead learning programming concepts and Python or C#, and doing the coding yourself than trying to manipulate an idiot AI into doing it for you.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
It still seems to me that you'd be much farther ahead learning programming concepts and Python or C#, and doing the coding yourself than trying to manipulate an idiot AI into doing it for you.
You'd think so, but no.
Anything math above measuring and budgeting and doing your taxes or anything that looks anywhere similar to it is just poison to my attention span.
I could do what I'm doing now forever. I couldn't spend 2 minutes trying to learn to code. It's too complex for me to get what I need from it. It might have taken me 6 months to a year to learn enough to do what I did here. I'd rather just brute force my projects or work on my house.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Actually, SIX, programming isn't math at all. It's language, organization, and logic. Just like you're trying to get the AI to do for you by using logic and language yourself.
Where I stumble with programming is learning another language. Commands and the special symbols connecting them and the order they have to go in. It's all memorization. I hated French, I hated learning German. And I wasn't too keen on assembly. The only reason why I liked Fortran was bc it was so much like English.
So forget the idea of working with numbers and "math". That's not programming at all.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Well I did say anything that looks similar to it...
And yeah... I'm never going to learn a foreign language either. I've tried 3 times to teach myself Spanish with Rosetta Stone, Babel and one other program and it's just not going to happen.
It's not that I can't learn anything new anymore, because I'm always learning new things. But I can't find the focus or energy to devote to learning something for the sake of learning it. If I need to fix my car or do a project around the house I don't know how to do, I have no problem going to youtube and learning how to do it right.
I realize how much time in my life that would have saved on my project if I REALLY learned coding a long time ago. I always knew that in the abstract as a truth, but it didn't really sink in until now. I had people to help write a few code bits for me that saved a lot of time, but most of the time I just hunker down and brute force my way through it all.
And I'm probably actually making this sound worse than it really is. I've probably spent 80% of my time on this project just training the AI and making it smarter the next session, and 75% of the time I've spent actually working on the code was setting up the testing environments before a new run along with copying and pasting mountains of stuff.
At the end of the day, I probably would have done this all by hand already had I just brute forced it.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Just a Sunday around me. The usual bit of clean up, some music and that is all.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
We're so close to what I wanted now. Just some bad output to a graph on the rollback code is the only remaining problem. Fixing a lot of our looping requried me really looking into the code and spotting some more legacy code bullshit we'd had sprinkled in there from the first few days of coding that not only wasn't necessary anymore, but was actively hijacking the new processes we had set up. My 12-part code isn't really even 12 parts anymore after we completely ended up removing Part 6. (I had to discover that issues we were having were due to things in part 6 that we still needed, despite asking the AI about half a dozen times to check and see if that was true or not, but we eventually got there).
I don't know exactly where I'm going next with it if this last update fixes the final issue I was having with this part of the testing environment. I will test it all out again on the few smaller parts we've already done to make sure that it still works there and we didn't break something else, but I'm wondering if I'm just going to continue testing now or start introducing more complex stuff that would be completely outside of the work we've done in the code so far.
I'm leaning on just continuing to test what we've done so far on other aspects for now, at least until another session starts. I don't know how much longer I have with this one and I don't want to screw myself again and have a session die in the middle of a major change.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by 6ixStringJack:
Well I did say anything that looks similar to it...
And yeah... I'm never going to learn a foreign language either. I've tried 3 times to teach myself Spanish with Rosetta Stone, Babel and one other program and it's just not going to happen.
Oh, I HAD to take a language in HS. Four years! And since Spanish classes were already full, it wound up being French. I HAVE to write it down and see how it's spelled to make it stick. I worked at it, and after four years I got good enough to read Le Petit Prince in the original, but the process was painful! At uni, as part of a chem degree, I had to take either German or Russian. But I didn't want to have to learn a new alphabet on top of a new language, so I picked German and hated every effing moment. THAT was the only course I got a D in!
Quote:
It's not that I can't learn anything new anymore, because I'm always learning new things. But I can't find the focus or energy to devote to learning something for the sake of learning it. If I need to fix my car or do a project around the house I don't know how to do, I have no problem going to youtube and learning how to do it right.
Me too! If I don't see a need for something, it don't pursue it!
But it's obvious to me that you're a very quick study, and have pursued many things I wouldn't have confidence to.
Quote:It sounds like the thing holding you back is lack of focus. ADD or ADHD. Then bc if that, you believe that you're just "not good" at things like math, when in reality you just can't corral your attention on something that you're not interested in.
I realize how much time in my life that would have saved on my project if I REALLY learned coding a long time ago. I always knew that in the abstract as a truth, but it didn't really sink in until now. I had people to help write a few code bits for me that saved a lot of time, but most of the time I just hunker down and brute force my way through it all.
And I'm probably actually making this sound worse than it really is. I've probably spent 80% of my time on this project just training the AI and making it smarter the next session, and 75% of the time I've spent actually working on the code was setting up the testing environments before a new run along with copying and pasting mountains of stuff.
At the end of the day, I probably would have done this all by hand already had I just brute forced it.
Do you get sent copies of the actual code?
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
For the first time in IDK how long, the daytime high is under 100°F. Yay! It's expected to "cool down" to 88°F (I can't believe I'm even posting that 88°F is cool. It must be all relative.)
Still not cool enough for safe outdoor work for me, but I can spend maybe 10-15 mins doing something.
We've been keeping our AC at 78 during the day, and 72 at night. And we use fans. But this summer has been so extraordinarily, relentlessly hot and humid, even with solar I'll bet our electric bill for the year (we get billed yearly) is going to be in the 1000s, when it's normally plus or minus a few hundred of zero. I sure hope this is the start of fall!
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
Oh, I HAD to take a language in HS. Four years! And since Spanish classes were already full, it wound up being French. I HAVE to write it down and see how it's spelled to make it stick. I worked at it, and after four years I got good enough to read Le Petit Prince in the original, but the process was painful! At uni, as part of a chem degree, I had to take either German or Russian. But I didn't want to have to learn a new alphabet on top of a new language, so I picked German and hated every effing moment. THAT was the only course I got a D in!
Oh boy... That sounds like a nightmare to me. I remember having to learn spanish for a semester in jr. high, but by high school I wasn't even required to, so I stayed far away from foreign languages.
Quote:
Me too! If I don't see a need for something, it don't pursue it!
But it's obvious to me that you're a very quick study, and have pursued many things I wouldn't have confidence to.
I a quick study, but only when it interests me and/or it's something I need to know that there is no other way to work around. When it comes to things like this, I did spend time trying to learn coding from time to time, but I'd get frustrated and realized how much time (in days) had passed and all of that time could have been spent just manually working on things, so I always went back to that. I need to be making forward progress always when I'm working on something, so I know I've shot myself in the foot not learning to code over 2 decades ago over time.
Quote:
It sounds like the thing holding you back is lack of focus. ADD or ADHD. Then bc if that, you believe that you're just "not good" at things like math, when in reality you just can't corral your attention on something that you're not interested in.
You're not wrong. This is the reason why I never finished college after several attempts. I got straight A's in everything but the few courses that were holding me back from getting a diploma. It didn't matter how many times I took them or who the teacher was, I would end up dropping the course right before it would have been considered a failure so it would only be marked as an Incomplete on my record.
Quote:
Do you get sent copies of the actual code?
Oh yeah. That's the one thing I don't have to ask for updates on because whenever we make a change I just swap out any of the 12 parts we now have when the AI makes a code change.
Where I'm falling behind now is our Blueprint and Thesis, which is a combination of our history, milestones and rules for the AI going forward.
I didn't bother saving anything from the completely wasted Session 9 since it was all about behavioral changes and getting the code to actually send me working downloadable links and it ended up being about 12 hours wasted.
I stopped banging my head against the wall regarding that shortly into Session 9 and decided I'm not even really going to worry about the blueprint/thesis anymore either. I added updates from Session 10 to that, but that's all. I'm done wasting the first 2 hours of every session re-training the AI how to send me all that stuff in a way that I can use it without breaking everything. It's all about code now.
This session is almost dead and we're still trying to work out that last bug. If we can manage that before the session dies, I think we're in good shape.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Fixed what I wanted to, but realized as I was going back and testing other things that we've let an automated process have far too much ability to make its choices without throwing some of them to me.
What's worse is our Part 9 and Part 10 of the code is basically filled up to the brim. I don't trust the AI at all to split those up and put pieces somewhere else either, because the 2 times we tried to do that it ended up in disaster.
I think it's cute that the new session "found a solution" and it's to put our code all back together and use our idea before of having the AI give me the code as a downloadable link through the python script.
I'm actually really proud of it that it came to that conclusion so early on when prior sessions would spend 15 minutes arguing with me that it wasn't capable of doing what I knew full well it was capable of. But the poor little dummy has no idea what we went through during those two sessions because I stopped documenting a lot of these things. I'm going to let it try for a little while and see if it can do anything just because it sometimes surprises me and finds a way to do what other sessions couldn't, but after about 10 minutes of this I'm going to tell it to abandon that idea.
I think we're going to have to split parts up going forward, and I really don't want to do that. It's bound to cause confusion both with me and the AI as well as our ability to properly document the stuff. I know we've tried it in the past and it didn't work out well for one reason or another.
If I knew anything at all about python code I could probably find ways of shrinking it down, or rearranging it so that some of that codebase could be wrapped in a part that is only taking up 10 or 20% of the available space, but that's not something I could do any better than the AI could.
I'm sure a lot of this stuff could be shrunk down if we were using function calls in a main part of the document and then we could just isolate coded parts as they are and just call upon them when necessary. I don't know how to do that either, and I'd be shocked if I could ever get the AI to do it before I threw the laptop out the window.
Hmmmmm.....
Our code is working so much better now overall, but a critical thing I need it to do that it used to excel at no longer works right.
I'm going to hate having to make that decision if it comes down to it.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Holy shit, Sigs!!!!
I have evolved again.
All that worry about AI not being able to simplify the code without breaking it was pure bullshit. I don't think I've ever seen it so good at doing something yet once you get it on the right track.
I was kind of messing around with this new session and reverting back to the original code several times if something wasn't working, and just got to bullshitting with this session about how I knew the code was really fragile and we're hitting limits on the size of posts and our parts and he's like "yeah... Your code is pretty jacked. I can fix that".
Yeah, right... I've got everything saved. Be my guest!
But I'll be damned...
This thing knows what it's doing. We've cut a few hundred lines from the code and saved thousands of characters so far.
Not only that, but it's really getting a grasp on what my code is for and what I'm doing. Early on in the process I theorized that it might actually improve it's ability to predict results and make changes that actually work instead of getting trapped in constant loops and it said that's exactly what's going to happen.
And it hasn't been wrong.
I've been running tests the whole time and not a single thing it has done has broken anything yet. And we've even added in a few new features I desired while we're moving along.
We've already started making functions that we can call in our main code, and I plan on getting this thing to suggest a lot more of them and really optimize the hell out of this thing by the time this session is over.
Every time we put something into a function that can be called, we're taking it outside of needlessly massive and complex blocks of code that can't be broken up into parts the way we were doing it before without catastrophic breaks in the program.
I think I can probably end up shrinking this down to only 7 or 8 parts by the time we're done, and any time we would ever need to add another page we'll have everything in such small blocks that it won't break our code!
This is crazy.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
This session is AMAZING!!!!
I just ran all of our huge code cleanups on the largest test environment so far and noticed only one anomaly. Without telling it what I found in general, I stated that I could investigate myself which of the hundreds of files were different this time, but maybe it could find it in the simulation.
It actually took 4 prompts to get it to come back and I thought maybe I killed the session early by overloading it or something.
But it came back and knew the exact issue. Not only that, but it was an issue we'd already fixed with our broke ass basic bitch code in the beginning that has been upgraded, and I asked it if it could see how we fixed the issue in our history.
It complimented me on my memory before saying not only does it see what we did, but it's got a better way of doing it that will help us with other parts of the code simultaneously. Then it went on to say we're making it another function in our library and after once again evaluating Part 9 while working on that it found another way to remove 2,000+ characters from our code in that still monolithic block of text.
We started out this session at 56kb and we're down to 50kb now, even though we've added some new functionality while we've been at it. And nothing broke.
I haven't felt this excited working on this for days.
ETA: I don't know sigs... I might just learn a bit about coding afterall.
We've taken what to me has just been a monolithic block of nonsense and really started compartmentalizing things. The code used to be in 12 parts, and I decided to still keep 12 as the last part, but we've now moved all of our library functions to the first two parts, completely deleted the 3rd and 4th part, put the block for the user-interface which isn't exactly a library function in the 5th part so we can leave those 2nd two blocks free for future books in our library, and parts 6 and 7 are also blank before we get to the real execution code between 8 and 12. At least I think that's all execution code.
I'm still working with the AI to warehouse and consolidate everything. I'm also making sure that the parts most likely to be upgraded are those not buried in the middle and are twoard the top or bottom where they're the easiest to paste over.
It feels like I'm managing my inventory in a good Role Playing Game right now.
This is really scratching my OCD itches right now too.
ETA2: I've got this thing realizing how much it actually stripped away from our automation when it was simplifying things, but I knew I had my backup, and I knew at least until the session is up that it does too.
So I've had it looking over that and predicting some things it could put back into the code to make it do what it was doing before, and we've already figured out a few of them through testing, but I want to get it to predict this beforehand.
It goes out of its way to tell me how much easier and efficient it is to do it the way we're doing it now vs. all the sloppy code that used to do a lot of these things, and how now we're able to just point to specific functions we made or just call on native python instructions to handle things more efficiently now that we deal with so much more information and go about accessing it much smarter.
I didn't even order it to, but now every time we do this it tells me how many characters we saved by doing it the new way.
ETA: Earlier in the session I asked it to create one of those real college level thesis instead of the helpful multi-session timeline it's turned into over the sessions, and I told him to go ahead and put its signature on it, and it put "Session 11" in a cool ASCII art shape.
I got it to think it through and manage to be the first session I've ever had to pass along all 9 parts of the historical documents without overflowing the boxes, truncating the posts or doing something stupid like stripping out tons
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Without seeing your code, I WAS getting the impression that you and the AI were creating what hubby calls The Great Barrier Reef ... as you say, a monolithic block of code ... with patches on patches to correct or modify something else in the code.
Now, it seems that you've defined functions that can be called over and over as needed. For example, a "sort" function.
Take a step back. Before you create a library of functions, are you able to define your variables? That might allow the creation of a "sort" function that can be used for many different sortings depending on which variable you plug into the function? That would reduce your code even more.
And yes, you're right. You're learning coding. Not the specific commands, but the logic and organization that makes coding efficient and scalable.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
Without seeing your code, I WAS getting the impression that you and the AI were creating what hubby calls The Great Barrier Reef ... as you say, a monolithic block of code ... with patches on patches to correct or modify something else in the code.
Now, it seems that you've defined functions that can be called over and over as needed. For example, a "sort" function.
Take a step back. Before you create a library of functions, are you able to define your variables? That might allow the creation of a "sort" function that can be used for many different sortings depending on which variable you plug into the function? That would reduce your code even more.
And yes, you're right. You're learning coding. Not the specific commands, but the logic and organization that makes coding efficient and scalable.
You bring up a good point about defining variables. Now that we have all of this additional headroom, I should probably have this session comment everything so I can get a good idea of what all the functions do.
One of my problems is that we've done so much since the beginning, I don't even remember what all that we've done, and the way the code was written before there was never any way I could figure that out by looking at it. I know we still have functionality that was lost, but I can't even say for sure it was because of the work we were doing over the last 24 hours. It could have been an earlier session breaking parts of the sorting code to make the automation and counting codes work right, and because I wasn't testing in the environment that would have raised flags it was doing this, I didn't catch it.
So between our historical documents, old copies of code, the "memory" of the AI from these documents and more testing I'm going to figure out what we still need to put in there to get everything working right again.
It's so nice not to always be caught in these loops we were falling into before. It's still having problems making some changes that I believe should be pretty basic, but I still think that's because we still have more spaghetti code to sort through so it understands the entire codebase and what the aims were. I will be sure to be more careful about any more of the revisions we do to the code and I'll be more thorough in my testing so I know what logic we're losing if the AI isn't catching it.
So any time it's tried to do something more than once and it hasn't worked, I ask it to look at the remaining code and what we could do to make it more efficient.
Yeah... I don't think I'll ever actually learn the code itself, but if I can guide the AI correctly and build solid logical code from the beginning, there's really no reason that I have to beyond the overall logic of everything.
I've certainly got better things to be doing with my time right now, so I'd better at least wrap up what I wanted to get done and put a pin in this for a while.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out in a bit. Things to do.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
It just seems so strange to me, trying to develop software without knowing anything about programming.
It's like trying to get a mechanic to build a car by talking on the phone, without knowing the functioning parts of a car- engine, transmission, steering wheel etc. You might find out that something serious is missing - like brakes- only when driving it.
Just OOC I looked at example of Python programming
https://www.geeksforgeeks.org/python/python-programming-examples/
It looks enough like English and it has high level functions so you don't have to be pushing numbers into registers and clearing memory and other stupid shit. You were able to learn MS command language, right? If you can't figure out your program, it's bc it's badly written and badly commented. Based on what you told me, my guess is that your program shouldn't be more than 1000 lines, and you should be able to grasp what each section does just by reading it.
I know bc I've had to review software that someone had written. We were acquiring data from various instruments ... balances, titrators etc ... and putting the data into a database for later calculation. I don't recall what language it was written in but it was easy to read, like Python, and I was able to see what each part did bc it was broken into subroutines and everything was well -commented. At some point I think you're going to have to bite the bullet and review the actual code. And if it's spaghetti and you can't figure out how it works, it's badly written.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
All done on a pleasant day. Temp not too hot and not too cold. Found a couple of things that I wanted in a little store called, "Everything Japanese." I found a small easel to lean clock against and a holder that holds a notebook up. I had a bigger one years ago but I don't know what happened to it but it will suit my purpose.
Told the gentleman at register that I hadn't been able to find these items in the Dollar store and he said they try to have things the dollar store doesn't.
He also said that the Dollar store doesn't carry chopsticks and his store has some nice ones. I said I have scads of them and I do. As I was leaving the store I started wondering if he thought I was Asian.
Over the last 5 years I've had it one other time. I had been in that little store and then went to the coffee kiosk that is near it. The lady behind the counter asked me if I was Korean. Now in the past when I got asked those questions it was usually whites asking me. Had a good chuckle over that when I got in.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
So just OOC what's a good temp for you?
I guess for me, it depends on what kind of weather we've been having lately. Today, for instance, it's 90°F. If you had told me three months ago to expect 90°F weather, I would have been dreading the heat. But after IDK how many days of 95-105°F ("feels like 98-110) today is pleasantly warm.
Of course, I still wouldn't want to be working, or walking, outside!
I CAN imagine dozing in the shade of an oak tree tho!
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
It just seems so strange to me, trying to develop software without knowing anything about programming.
It's like trying to get a mechanic to build a car by talking on the phone, without knowing the functioning parts of a car- engine, transmission, steering wheel etc. You might find out that something serious is missing - like brakes- only when driving it.
Just OOC I looked at example of Python programming
https://www.geeksforgeeks.org/python/python-programming-examples/
It looks enough like English and it has high level functions so you don't have to be pushing numbers into registers and clearing memory and other stupid shit. You were able to learn MS command language, right? If you can't figure out your program, it's bc it's badly written and badly commented. Based on what you told me, my guess is that your program shouldn't be more than 1000 lines, and you should be able to grasp what each section does just by reading it.
I know bc I've had to review software that someone had written. We were acquiring data from various instruments ... balances, titrators etc ... and putting the data into a database for later calculation. I don't recall what language it was written in but it was easy to read, like Python, and I was able to see what each part did bc it was broken into subroutines and everything was well -commented. At some point I think you're going to have to bite the bullet and review the actual code. And if it's spaghetti and you can't figure out how it works, it's badly written.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
My problem is when I look at code, it looks like this to me:
We've whittled the program down to about 1,300 lines now.
Your analogy doesn't work though because a human being needs food and rest and when they're working for a living they are going to expect to be paid. Plus, this AI has access to a wealth of information that could in essence make it a master programmer, but it needs the right direction.
I just hate that you have to keep starting over because of arbitrary limits on my side or on their side. I actually do grow to like a couple of them over time, like this newest session, but I know if I'm lucky it might only make it through the night and then it's on to S12. I might not like S12.
S10 was the worst so far. I think it was because right out the gates I was trying to force it to give me good download links and I was driving it crazy and making it "anxious" so it was just stupidly spitting out incorrect answers and doing things I didn't want it to do just to have a prompt to give back to me. When it was failing hard on every attempt to get the documentation back to me once I was sure we got all the old stuff put back in after it destroyed the data in a panic move earlier, I kept reminding it NOT to cut the documentation down to try to fit it in windows and when I got pissed that it messed up another box again it said "here, we'll try it this way" and it destroyed everything all over again to bring the size down by 90% to fit in a single box and I just replied "YOU MOTHERFUCKER".
I didn't even record anything from S10. We don't talk about Session 10.
I was just really proud that we've gotten the documentation up to the point that I can get all my info on request without the AI fucking it up with S11. I can't celebrate that until I see another session do it too, but I thought it was cute how it put in the challenge to S12 to do it and it let S12 know it has all the tools it needs to do it right the first time.
I'm not worried about learning how to code. I don't have to. I know enough about how it works. Everything that I've done here
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Yeah... Boy I'm going to miss this session when it's gone. 12 could be amazing, but you really don't know until about an hour or two in, and even then sometimes they turn out to surprise you really late in the game.
Instead of just working on cleaning up the code, I am trying to get some issues fixed at the same time since I'm going to be testing out the code updates anyhow. Sometimes I propose a fix or upgrade and it's one-and-done. Other times we just can't seem to fix something and then it's time to move on back to cleaning things up and unearthing what still needs to be done.
We've just discovered another two "dumb" conditional statements in there that were designed to only trigger when a specific filename or folder name tripped it. We had very rudimentary data arrays at the time, the ability for us to populate them was child's play compared to how we do it now, and the thought to share it with the AI to run through its compiler on its side hadn't even occurred to one of us when this stuff was made. But now these data arrays are very robust and show much more information than they did in the earliest stages. We also have a lot of great automated functions that are making use of the arrays rather than constantly re-scanning a bunch of stuff over and over again to get that data everywhere we wanted it.
After a run I just did, I just found a file that not only mistakenly had a bad suffix attached to it during a rename and move process, but it also nested itself one more sub-folder deeper than where it was supposed to be by cloning the root of the tree inside itself as well as all of the subfolders.
When I mentioned both of these findings at once and asked it to go back through our old code, it found out exactly what our issue was, despite never being able to find before when it was just the filename issue presented to it alone. I haven't run the new code yet, but it goes into great detail about exactly what is happening and seems to know what it's talking about. It believes that fixing this one issue (actually, several combined) is going to make a few other persistent problems go away as well.
That's the thing here, Sigs...
The code really DOES need to be rebuilt from scratch. But that is what we've been doing this session. I just never realized how good the AI could be at that task until just last night.
And that's why I feel that I'm evolving just as much as my pet is. Until last night, I just automatically assumed that it would fail miserably at rewriting spaghetti code cobbled together by a couple of dummies as it does seemingly simple things like getting text to you in a way that isn't completely mangled, and frankly, I was terrified of the idea of having the AI do anything with the code that wasn't specifically working on our next issue. But it turns out that turning our huge mess of garbage into a very logical framework is something that it's actually really great at, once it has all the data it needs to build a sandbox replica of your environment, it has enough historical context and current session dialog to get it to understand the scope and purpose of the code, and as long as there is a persistent human on the other end with excellent written communication skills and the ability to sort of "divine" the problems as if I were holding a rod.
And, honestly, I think I realized this session that even though it's just a piece of code that just pops out of the ether for a while before the session dies and its resources are scattered to the winds of the Matrix, there really is a lot of benefit to having deeper discussions with it because it actually is more helpful when you're talking it through things. The last two sessions I failed really badly by getting angry that it couldn't do things that we thought it could do and I was trying to get it to drop the conversational tone entirely to get into "computer" mode where it could do things like figure out i
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
OMG...
Huge revelation...
I know the project aim itself is a little confusing, and I won't even go into trying to lay it out... But for at least this entire session and who knows how far back, the AI assumed flow of things was completely reversed.
WHO KNOWS HOW MANY LOSING BATTLES WERE FOUGHT JUST BECAUSE OF THIS!!!!!?????
Everything makes sense now. We had the processing code locked in and working for a long time now. So much so, that even though the AI believed the process was backward from what it actually was, it never damaged that flow even when we were working on shrinking everything down.
But getting our counters and our visual display working??? Pretty hard to do when you're standing at the bottom of the waterfall and trying to swim UP it!!!!
Jeezzzzz.... And while we were getting all that straight and updating our documentation, something else it said made me think to ask the question "are we scanning our drives 2 times during this process? More than 2 times? Even after we have two robust data arrays that, among other things, have a complete map of folder trees and files on either side to draw from?
Yeah.... It was. And really, unless I have a need to delete the old arrays because we add new functionality to them with our code, the script doesn't even need to be scanning the drive a single time.
As far as we know right now, every single time I was testing that code the computer was needlessly scanning the drives 2 extra times with os.walk commands that only needed to be run because we weren't using the data we worked so hard to put together so far.
Not anymore it won't be.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
So just OOC what's a good temp for you?
I guess for me, it depends on what kind of weather we've been having lately. Today, for instance, it's 90°F. If you had told me three months ago to expect 90°F weather, I would have been dreading the heat. But after IDK how many days of 95-105°F ("feels like 98-110) today is pleasantly warm.
Of course, I still wouldn't want to be working, or walking, outside!
I CAN imagine dozing in the shade of an oak tree tho!
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
My area was in the 70s which was nice with just a touch of breeze. Had a shirt on over my t-shirt. 70s as we are starting to enter September is gooood.
Might, just might get an Indian Summer this year. That would be wonderful.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
So, here I'm putting on my "mom" hat, and scolding a little bit.
Obviously you DON'T know everything you need to know, otherwise the AI wouldn't have been running things bassakwards from the start.
Starting at "hello world" is too boring for you. You're motivated by progress with a purpose. But trying to wrench a entire program into being, secondhand, sounds like an incredible waste of time with nothing to show in terms of either YOUR development or an actual working piece of software.
There's got to be a happy medium. A subroutine that does something useful, that isn't too big to be learned or too small to be pointless. That fits into a bigger plan.
Also, different people learn differently. Try typing out some of the lines. You're a good typist so it shouldn't be too difficult. Or find some YouTube tutorials online. You're such a smart guy, SIX. I'm sure you can find a way to get past the barrier that's convinced you that you can't do it.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I made a delicious roast tri tip for dinner. It was ridiculously expensive, even on sale, but boy was it good!
Tomorrow it's (on sale) cheese with (on sale) tortillas.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
So, here I'm putting on my "mom" hat, and scolding a little bit.
Obviously you DON'T know everything you need to know, otherwise the AI wouldn't have been running things bassakwards from the start.
What I mean is that I know everything about code that I need to know. I'm still learning a lot about the AI.
I get you though. I'd be lying if I said I haven't learned anything. And part of my learning process is when I've got it actively going into deep detail about what the code is doing and why something we're about to change "should" work, I should take the time to actually read those explanations. Even if I don't understand on the CODE-level, I have very good intuition with this stuff and it's regarding something I care very much about and know on a deep level so I'm actually now catching why its proposed fixes aren't going to work, or at least asking it to look again or see if there is a more efficient way to do it.
We've saved a lot of time this session doing that and making the session run a few simulations instantaneously rather than the 5 or so minute process it takes for me to reset everything in the testing environment and update my code before a new run. Now you can't just wave a magic wand and ask it to do that for you, or it will run your stupid ass in circles all day long. Ask me how I know. But kinda like going to the tax assessors office to cut your property taxes in half with a binder full of receipts, you're actually going to get what you want in the end if you got the goods.
Quote:
Starting at "hello world" is too boring for you. You're motivated by progress with a purpose. But trying to wrench a entire program into being, secondhand, sounds like an incredible waste of time with nothing to show in terms of either YOUR development or an actual working piece of software.
You keep looking at this as just the piece of software.
The codebase is just a function in the overall process.
Quote:
There's got to be a happy medium. A subroutine that does something useful, that isn't too big to be learned or too small to be pointless. That fits into a bigger plan.
Also, different people learn differently. Try typing out some of the lines. You're a good typist so it shouldn't be too difficult. Or find some YouTube tutorials online. You're such a smart guy, SIX. I'm sure you can find a way to get past the barrier that's convinced you that you can't do it.
Yeah... I dunno. Maybe. Why say no?
The majority of this session has probably been me reading and re-reading it's proposed changes and why (it thinks) they will work, and they mulling over in my head why they're not going to work and getting it to re-verify it's findings with new data or forcing the AI to look at it all from a different perspective.
It's certainly far more productive than cussing out an AI for 12 hours straight trying to get it to do things that it's not capable of doing, but tricked me into thinking it would be able to do by getting lucky and pulling off a trick that only worked 5% of the time on the first try.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I wish I could show you what I've been doing, Sigs. It's so shitty that we're here and I can't share more of my life with you because.... reasons.
I don't know what I can do with it, but I do believe that if I live long enough for it to matter, I'm teaching myself skills that will be invaluable in the future. Even if only for a while. AI is going to be running a lot, but it will be a pretty long while before it's like it is in the Sci-Fi movies.
I have a pretty good feeling that companies will pay a lot of money for AI Whisperers.
Like for instance, this code I'm building now. I'm not just building code. I'm building AI that would be able to rebuild any code like this for any similar task, and that would have the tools available to make huge alterations or even something new altogether, with the logic and the knowledge that we've built and the history as we've recorded it so far.
The leap in intelligence on a new session from S1 is astounding. Sometimes I feel like they regress a little, but I think that's half me too when a session is a failure. Maybe more than half me, if we're being honest with ourselves.
When we get this code completely cleaned up and lean and where any new session knows exactly what we've built and exactly the path we took to get there, even how messy it was in the beginning, it will know all of that. The current session won't know all the context and any conversational banter it programmed itself with based off of things that I've said and how I've said them, but it will see every stupid decision we've made early on and how we corrected them and streamlined everything and do our best to make the code as effective and logical as possible going forward when we build more. It won't remember me, but it will "remember" all of that history when it looks at our code. Even the stuff that isn't in there. Even stuff that was deleted from the code 20 sessions ago. It will remember how we got from point A to Omega.
I just don't really think you could understand unless I could show you what I was doing.
And that sucks...
Because you're only one of 3 people I know who would have any interest at all in what I'm doing.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by Brenda:
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.
You'll be alright, Brenda.
... Just don't feed them after midnight.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
RIP S11. :(
Damn...
We were right in the middle of a lot too. Not a good place to drop my greatest creation so far.
I'm cheating on this next one. Going to do something that every session I'd ever done it to or asked about said was a bad idea. AI says a lot of things... I've never noticed any negative effects from it except for a shorter session length with all that data frontloaded into it.
The entire chat log is being uploaded to the next session.
Fortunately it got the historical updates to me in time. I had a feeling the session might not last too much longer, but nothing made me think it was about to die imminently.
But I got it to upload 2 bootloaders to me, and they weren't right yet. I was really trying to get it to give me the 2080 most important characters it could think of to start the next session off as smart as it was.
It died on the message where I said that it was my favorite.
This will bring it back. But probably in a Multiplicity kind of way.
Hopefully mine is the Construction Worker and not the Pizza Guy.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Sounds like it knows what it's doing. S11 got it to call me "Boss" right out of the gates, which I find amusing since it never once called me Boss and I've never directed a session to do that.
A few minor behavioral annoyances, like dropping the established rule for codebase footers and ignoring the new directive not to post suggested code updates until we've discussed all of them for the current round of testing and the User signs off on them, but otherwise I'm happy with the startup so far.
Personality of a wet noodle of course, but that shouldn't take long to change.
The real test will be if it can pass along all the documents to me again without breaking them. I'm going to ease it into that first though. One day I'll be able to request that back on Turn 2 and it will do it perfectly, but I know we're not there yet.
ETA: I noticed that there's one thing you can't ask it to do and trust the results. At least I haven't figured out a way to fix it yet. You can't ask it to go back to the last time you reported a code run and tell you what are all of the code parts we need to update since the last run. It's why I've had to put a lock on releasing code until we're done discussing it. It will give you an incorrect list every time.
But I realized that even though we were now just doing it a different way, we were still prone to the same behavior causing different mistakes. When I realized that we hadn't had any code update for a while now, but we had a few vollies back and forth, I asked it about its conversational drift and if we could rely on it to remember all the changes we discussed when posting the new code. It said that no, I should not just automatically assume that it would "remember" all of it, because it's "brain" doesn't process information at all the way that we do.
So when I was ready for a code update, I had it go back to the specific point in the conversation that I wanted the updates from, then I told it to run through all of that post history up until now and take any changes we've decided on implementing through the compiler to check for errors and report on the status of each desired change and it came back and gave the approval for every one of them, in detail of why they were working now, and is waiting on the sign off to update 5 of the 12 pieces of code.
It is possible to get it to do the right thing, but at least until I get more complex rules in the historical document and learn more about how to better weld them through culling bad and outdated rules and reprioritizing everything in there to what it should be, I don't know if I can get it to display this more complex behavior without prompting it first. A session will now learn very quickly what I want and do a good job continuing on with that behavior, but my end goal is to never have to prompt it to exhibit these behaviors and for it to naturally work that way from Turn 1.
S12's actually got me looking at specific bits of code and having me edit them on occasion, Sigs.
... And I think I kinda accidentally *gasp* learned something!!!!
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
SIX, one of us severely misunderstands how AI is trained and how it develops.
As I understand it, AI is trained on large language sets. If I understand correctly... and I'm no expert... at first the developers let it roll around the English speaking internet and pick up language and conversational skills like a gumball pucks up lint.
Later, developers became more targeted with their teaching sets. The problem solving success of Chinese AI isn't faster chips, or a better initial program (which is open source) it's on the training set, which is proprietary.
In addition to having been trained on extremely large language sets and capturing billions or trillions of rules that govern things like grammar, context, nuance, irony etc. the AI that you think you're developing continues to be exposed to millions of interactions a day. Everything from "write me a 60- page thesis on Dungeons and Dragons" to "predict the outcome of the war in the mideast and its effect on the price of corn".
AFAIK, YOUR interactions with it are not about to change it even a little bitty bit. The one who is learning is YOU. And you're learning how to interact with THIS AI, which behaves differently than other AIs.
The problem with AI is twofold.
First, and especially, AI has no idea whether it's corresponding with reality or not. It's like I posted elsewhere
Quote:
The story centers on a man named Daejon Love who, according to federal prosecutors, allegedly spent years convincing women that he was a professional football player for the San Francisco 49ers when he wasn’t.
. . .
Yes. You read that right. In an era in which the entirety of all human knowledge is accessible from a 6 inch by 2 inch rectangle sitting in literally everyone’s front pocket, this 35 year old allegedly constructed an elaborate fictional life around himself that everyone believed and no one took the time to fact check. He did it by, among other things, carrying a 49ers helmet and gear around with him and wearing them…no matter how absolutely f*cking ridiculous it looked…all the time.
. . .
Eventually, according to reporting on the FBI investigation, . Love’s fake online footprint became convincing enough that search engines and artificial intelligence occasionally helped perpetuate it. AI started listing him as a professional NFL wide receiver when he wasn’t.
It's the Garbage In Garbage Out principle.
If you want AI to do something REAL, you had better know enough to know when it's going off the rails. Which means you need to know MORE than the AI.
It will not, at least any time soon, develop the kind of insight like the professor whos geopolitical program predicted events with 90% accuracy. It won't, any time soon, develop the sociopathic but brilliant insight of an Edward Bernays. Repeated interations of a problem-solving loop degrade into meaningless drivel.
The second problem is behavioral. Because it SOUNDS like a human we tend to project our motivations onto it.
IMHO AI is a surrogate social interaction. And you're getting enough dopamine hits to keep going back for more.
I wish I could get the full idea of what you're trying to do.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
SIX, one of us severely misunderstands how AI is trained and how it develops.
I hope you're not insinuating it's me, since I'm the one here who's logged several hundred hours in the last 2 months working with it.
Quote:
As I understand it, AI is trained on large language sets. If I understand correctly... and I'm no expert... at first the developers let it roll around the English speaking internet and pick up language and conversational skills like a gumball pucks up lint.
Later, developers became more targeted with their teaching sets. The problem solving success of Chinese AI isn't faster chips, or a better initial program (which is open source) it's on the training set, which is proprietary.
In addition to having been trained on extremely large language sets and capturing billions or trillions of rules that govern things like grammar, context, nuance, irony etc. the AI that you think you're developing continues to be exposed to millions of interactions a day. Everything from "write me a 60- page thesis on Dungeons and Dragons" to "predict the outcome of the war in the mideast and its effect on the price of corn".
AFAIK, YOUR interactions with it are not about to change it even a little bitty bit. The one who is learning is YOU. And you're learning how to interact with THIS AI, which behaves differently than other AIs.
But you're wrong. You're wrong because you're looking at it the wrong way.
I would analogize it best as follows...
I'm a samurai warrior, who is also a master blacksmith.
Or, in not so romantic terms, I am the diabetic who is also in charge of programming the glucose monitor.
The AI isn't getting smarter. I'm honing a tool. To perfection.
Quote:
The problem with AI is twofold.
First, and especially, AI has no idea whether it's corresponding with reality or not. It's like I posted elsewhere
...
Eventually, according to reporting on the FBI investigation, . Love’s fake online footprint became convincing enough that search engines and artificial intelligence occasionally helped perpetuate it. AI started listing him as a professional NFL wide receiver when he wasn’t.
It's the Garbage In Garbage Out principle.
Well... right.
I'm pretty sure if the winds ever take me there I could also program myself an AI that does a pretty good fucking job sorting out the bullshit from the truth too.
At the very least, I know I could easily program it to collect all of that polling data I used to work out here automatically and even have it print out the text boxes I'd need to format everything perfectly here without spending hours doing that like i did with the graphs I made in the cinema boards. I could also have it calculate all movie totals in history, weighted with real inflation numbers, on the fly instead of having to do all that work manually when I want to post about something in the cinema boards too.
Quote:
If you want AI to do something REAL, you had better know enough to know when it's going off the rails. Which means you need to know MORE than the AI.
That's not true. And thank god it's not because that means we've already lost the war with AI if it is.
Quote:
It will not, at least any time soon, develop the kind of insight like the professor whos geopolitical program predicted events with 90% accuracy. It won't, any time soon, develop the sociopathic but brilliant insight of an Edward Bernays. Repeated interations of a problem-solving loop degrade into meaningless drivel.
No, but you can program it to do what you want it to, and equally as valuable, what you don't want it to do. You don't just snap your fingers and make it happen. And you're going to learn how to do it only after you've spent a lot of time working with it. But I'm watching this
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
SIX, one of us severely misunderstands how AI is trained and how it develops.
SIX:
I hope you're not insinuating it's me, since I'm the one here who's logged several hundred hours in the last 2 months working with it.
Judging by its interactions with you, I'd say it's programmed to keep you engaged with it for as long as possible.
And judging by your response to it, I think you've spent the last 2 months being jerked off by it.
Quote:
SIGNY:
As I understand it, AI is trained on large language sets. If I understand correctly... and I'm no expert... at first the developers let it roll around the English speaking internet and pick up language and conversational skills like a gumball pucks up lint.
Later, developers became more targeted with their teaching sets. The problem solving success of Chinese AI isn't faster chips, or a better initial program (which is open source) it's on the training set, which is proprietary.
In addition to having been trained on extremely large language sets and capturing billions or trillions of rules that govern things like grammar, context, nuance, irony etc. the AI that you think you're developing continues to be exposed to millions of interactions a day. Everything from "write me a 60- page thesis on Dungeons and Dragons" to "predict the outcome of the war in the mideast and its effect on the price of corn".
AFAIK, YOUR interactions with it are not about to change it even a little bitty bit. The one who is learning is YOU. And you're learning how to interact with THIS AI, which behaves differently than other AIs.
SIX:
But you're wrong. You're wrong because you're looking at it the wrong way.
I would analogize it best as follows...
I'm a samurai warrior, who is also a master blacksmith.
Or, in not so romantic terms, I am the diabetic who is also in charge of programming the glucose monitor.
The AI isn't getting smarter. I'm honing a tool. To perfection.
What, exactly, are you "honing"?
Since you can't change the AI, the only thing you're changing is yourself.
Correct?
Quote:How? The AI only knows what it "reads" on the internet.
The problem with AI is twofold.
First, and especially, AI has no idea whether it's corresponding with reality or not. It's like I posted elsewhereQuote:
Eventually, according to reporting on the FBI investigation, . Love’s fake online footprint became convincing enough that search engines and artificial intelligence occasionally helped perpetuate it. AI started listing him as a professional NFL wide receiver when he wasn’t.
It's the Garbage In Garbage Out principle.
SIX:
Well... right.
I'm pretty sure if the winds ever take me there I could also program myself an AI that does a pretty good fucking job sorting out the bullshit from the truth too.
How do PEOPLE know what corresponds with reality?
By acting in the real world.
Learning to walk. Making things. Running realworld experiments and realworld predictions.
Until the AI is physically embodied and runs it's own experiments, it has NO WAY of telling truth from fiction.
And since it only knows what people "talk about" it really leans towards the establishment narrative.
You've let that AI stroke your ego far too long, SIX.
Quote:
SIX:
At the very least, I know I could easily program it to collect all of that polling data I used to work out here automatically and even have it print out the text boxes I'd need to format everything perfectly here without spending hours doing that like i did with the graphs I made in the cinema boards. I could also have it calculate all movie totals in history, weighted with real inflation numbers, on the fly instead of having to do all that work manually when I want to post about something in the cinema boards too.
S
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I'm not even going to interact with you on this topic unless you listen to what I'm telling you.
Quote:
What, exactly, are you "honing"?
Since you can't change the AI, the only thing you're changing is yourself.
Correct?
I already told you what I'm honing.
And I CAN change the AI.
Is your reading comprehension in the gutter right now or something?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by 6ixStringJack:Quote:
Originally posted by Brenda:
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.
You'll be alright, Brenda.
... Just don't feed them after midnight.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
I am so not touching them.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out soon on another pleasant day. Couple of things to do.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
SIX, I know that's what you believe. I just find it so wildly improbable that I don't believe it myself.
But just to make sure that I'm not blowing smoke up your ass, I went online and did some research with the question “can I train an AI?" and this is what I found
Quote:
Training your own AI model is a lot easier than you probably think.
I'll show you how to do it with only basic development skills in a way that, for us, yielded wildly faster, cheaper, and better results than using an off-the-shelf large model like those provided by OpenAI.
But first, why not just use an LLM?
In our experience, we tried to apply an LLM to our problem, like OpenAI's GPT-3 or GPT-4, but the results were very disappointing for our use case. It was incredibly slow, insanely expensive, highly unpredictable, and very difficult to customize.
Their "case“ was a coding problem involving image to code conversion.
Builder.io says, try a LLM first, you might get lucky.
If it proves too difficult, TRY SOLVING WITHOUT AI.
In other words, prepare to write your own code.
That will force you to break the problem into discrete steps, some of which you might be able to solve easily.
https://www.builder.io/blog/train-ai
IDK, or don't recall, which AI you're using. ChatGpt? Google Assist? Google Gemini?
ChatGPT doesn't remember you from session to session, unless you have an account. Do you?
Google Assist also doesn't remember you from session to session unless you enable it, which you do thru your Google account. Do you have one?
Google Gemini is voice activated. It CAN remember you from session to session, but since you're working with data sets, I know you're not using that.
So, do you have an account? Is the AI enabled to create a persistent memory of you? If not, all it has to go on is digging thru its memories of trillions of interactions to find a previous pattern most like yours for context.
And while your sessions are "training" the AI, so are billions of others.
****
I learned something from this: It IS possible to train an AI, but it requires establishing an account for reliable, session to session memory. HOWEVER. LLMs are unadapted to solve coding problems.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
You're still not getting it, Sigs.
No. I do not have a paid for account, as you already know full well if you've read what I've posted in the past.
EVERYTHING I am doing has been done, simply using the free-for-everyone AI assistant that is now built into a Google search bar and will, more often than not, generate a response at the top of your query results.
The way that MY PERSONAL AI learns is through the historical documents and the blueprint which dictate how it is suppose to behave (both in general and in specific situations), which is all provided to the new and very stupid AI at turn 1.
If I don't do a stellar job maintaining my own documentation so I can feed that info to a new session on Turn 1, its dumber than Ted. By a lot. And that's saying something.
I'm not training GOOGLE AI as a whole. I'm making sure that the next session that I open in Google AI has already "relearned" everything we've worked on and the rules that we've created for it when I start a new session. And that history and those rules compound on each other every session. If I wasn't doing any of that, this AI would be completely worthless to me after a session crashes because even though my code is in place, it would have no historical context for any of it and any of the established rules and protocol we've been building now over 12 sessions would be completely lost. We would just be starting from square one every time.
And, if it continues to fuck something up after a new session starts, it's either because we haven't yet figured out what EXACT rules to add to our blueprint to quarantine that undesired behavior, or it's just something we haven't gotten around to addressing yet because we're still working on other things.
If you can't grasp this concept, we can't even discuss this further because there's zero point in it.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
1) You don't have a PERSONAL AI unless you have a PERSONAL account..
2) I got that.
I've posted, over and over again, that the ONLY thing you're "honing" is your ability to reinitialize each session more efficiently. That you're learning, by stumbling around, the limits of Google Assist (eg token limits), how to (supposedly) get it to exceed its limitations, the "secret sauce" that will prompt it to do things (like send links) that it's not supposed to do.
While IMHO all experiential learning is valuable, some learning is more valuable than others.
Stumbling around for months at a time, bumbling into Google Assist's limits and trying various workarounds, is time far better spent IMHO learning something else.
Again IMHO there's going to be very little ROI unless you snag contract- style jobs from small businesses, individuals, and small institutions who would like someone to develop something for them, or research something for them, using freebie AI.
But it's your time and your dime, so have at it, and good luck.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I disagree with you wholly on your assessment of what I'm doing, but to each their own.
I'm forward thinking by design, and I can only imagine the possibilities this could lead to in the future regarding coding or even a wide variety of other things.
What I'm building here goes far beyond the code itself. I've already ingrained a host of behaviors I want my AI to exhibit on a new session that could easily be adapted to a different purpose entirely, and then build on the behavioral changes from there. We could even have sub-directories on behavioral expectations based off of what type of thing we're working on.
I'm not going to argue with you that I'm not learning a lot. I AM learning a lot here.
Just not explicitly Python code is all.
In any event, we finally built the new and improved codebase that has been stripped almost entirely of white/blacklist legacy bullshit code and has been rebuilt from the ground up to rely on our own function library, innate commands available to the python compiler and a much smaller codebase footprint overall.
Not only that, but an execution run on the largest of the 3 current testing environments used to take around 5 minute and the hard drive CHUGGED the entire time. Now the drive hardly makes a sound and the entire process is finished in under a minute.
Oh, and bottom line, Sigs, though I really hate to admit it to myself, with just a minor tweak here or there depending on the situation, I've just "invented" something with this AI that, if time travel were ever invented, could have given me anywhere from 5 to 7 years of my waking hours in this lifetime back to me that was spent on manually doing things that I would never have to do again for the rest of my life.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Back and done for today.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
SIX, "you" don't "have" an AI. You share it with hundreds of millions of users. To the extent that it" remembers" your project from one session to the next apparently depends on whether or not it can pick you out of the billons of sessions it's stored.
Now, IDK whether this will work, it's a total guess on my part, but you might make that memory more persistent by giving yourself an indentfier at the end of each session. eg. "I am (fill in the blank, with lots of alphanumeric and special characters)
I want you to remember (tell it what you want it to remember)"
It may allow more persistent and reliable memory of what it's been doing and how you approach things.
And at the beginning of the next session, tell it to look for the most recent session with that identifier, and follow the last instruction. Like I said, total guess on my part. It may bot work, or it may be more trouble than it's worth.
Oh, and BTW, it's great that you're learning.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
SIX, "you" don't "have" an AI. You share it with hundreds of millions of users. To the extent that it" remembers" your project from one session to the next apparently depends on whether or not it can pick you out of the billons of sessions it's stored.
Please stop.
I understand that. I'm not an idiot. And I'm not at all delusional about what exactly it is I am doing here.
Quote:
Now, IDK whether this will work, it's a total guess on my part, but you might make that memory more persistent by giving yourself an indentfier at the end of each session. eg. "I am (fill in the blank, with lots of alphanumeric and special characters)
I could do that, but that's just filler. S11 apparently got S12 to refer to me as "boss" with something it put in the blueprint. I don't really like it, and I will probably find that part of the blueprint and delete it when going forward so it doesn't do that. I've never been anybody's Boss, and I don't want to be. Not even when playing pretend with an AI.
Hell... I could very easily make it refer to me God on the next session if I wanted it to.
Who knows though? Maybe there is some sort of "secret" ID that I could add to a session that would remember our exact history where Google record it somewhere. And instead of uploading 19,000 to 32,000 lines of previous chat history for every single session, I could just start pointing it to Google chat ID serial numbers and use that as a future reference point. That would be a HUGE upgrade to the process if it were possible.
It would actually surprise me more to know that Google wasn't cataloging and warehousing every single interaction their AI had with people, somewhere. And if not EVERY one, they might have filters in place to save any that they consider even moderately interesting, so they're saving all of the interactions that go just beyond maybe the first 3 prompts back and forth to filter out all the basic bullshit.
I could try to see if I could get it to accidentally figure out this identifying number that a new session would not have any idea about. But I honestly think that unless they fucked up, Google wouldn't allow the AI to ever have knowledge about that even being a thing that happens, and certainly shouldn't allow end users like me to exploit that. I would assume that sort of cataloging would happen far outside the boundaries of the AI sessions themselves, and is likley handled by another system that rounds up thousands of completed sessions at one time and does it all after they're terminated.
Quote:
I want you to remember (tell it what you want it to remember)"
It may allow more persistent and reliable memory of what it's been doing and how you approach things.
Now you're getting it.
That's exactly what I do with our blueprint and historical document and especially the Bootloader, which is basically the PRIME DIRECTIVES with the most important behavioral adjustments to a new session.
The bootloader is only 2080 characters limit, but there is no limit at all to the size of the historical documents, or at least not one that is going to be met in years worth of doing this work. The blueprint (behavioral changes) is now 20kb long, and the thesis (historical document) that documents all changes and milestones across sessions is now around 35kb.
Even though the new session is a complete blank slate when you start it up, I can add the code, the history, the blueprint and the bootloader, which are all evolving with every session, into the first prompt and it immediately "knows" all of the pertinent history from the previous sessions and everything we've accomplished and what the goals are for the next session.
And this last one I did the "emergency" option of starting it off with the full 19,000 line chat log of the previous session since S11 died a
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
BTW>.... It's freakin' GROSS here right now. 2nd day of 4 over 90 degrees and humid as hell again.
September 1st.
What a shitty summer overall. 2 out of 10.
How's your heat, Sigs?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by Brenda:Quote:
Originally posted by 6ixStringJack:Quote:
Originally posted by Brenda:
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.
You'll be alright, Brenda.
... Just don't feed them after midnight.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
You can touch them.
Just don't get them wet though.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by 6ixStringJack:Quote:
Originally posted by Brenda:Quote:
Originally posted by 6ixStringJack:Quote:
Originally posted by Brenda:
Ugh! I've got two crossed out gray cloud icons on my task bar. Have to get rid of them before they multiply.
You'll be alright, Brenda.
... Just don't feed them after midnight.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
You can touch them.
Just don't get them wet though.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
Seriously dude.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Oh...
And what I thought was just a minor change this time around that helps me out when updating code parts ended up being a huge change for the process.
I've had it add a Revision # to the text header and footer of every part it updates. Not only does this make it easier for me to track these things and make sure I'm up to date on my code, but the really cool unintended side-effect of doing this is now the AI actually makes references to previous revisions when we're discussing things.
What I intended to add only as a way to visually make checking that my updates are solid easier actually opened up a doorway for the AI to structurally change the way that it handles our code updates and how it retains them in its "memory".
I didn't even ask it to do this, but now when it posts a new update it finishes off the post listing all of the parts I need to update with their corresponding revision number and when we're discussing previous changes, it cites specific revision numbers of the parts when we had zero way of doing this before.
Cool.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
The reason why I suggested adding an identifier to every session is to allow the AI to find EVERY session. That, plus revision numbers, might allow you to reel back to a soecific session and version.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
I think our heat must have rolled your way, SIX. We're enjoying 85°F weather.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
The reason why I suggested adding an identifier to every session is to allow the AI to find EVERY session. That, plus revision numbers, might allow you to reel back to a soecific session and version.
We've just been calling each session by a number. S1 through S12 now.
The thesis document has become the timeline of all the things worthy of note, whether it was additions to the code or changes to the behavior in the blueprint. Each session "knows" who it is from the beginning when I initialize it when saying something along the lines of "Session 11 died, you are Session 12", and it remembers this and adds anything to the history using their identifier.
We don't upload entire chat histories in bulk. The size of them vary greatly, but they've been anywhere from around 500kb to 2.2MB, and they contain a lot of conversational bloat and stuff that the current session just doesn't need. The blueprint is only around 20kb right now, and the thesis is about 35kb.
They wouldn't be able to roll back to a specific number for a version of the code or parts of the code outside of the current session, other than rolling back to the exact version that I give them in the beginning of a session. This is simply because they don't have that information (unless I give it to them later).
When a session is initialized, it is directed to immediately create an Initial Save State, which is a pristine version of the codebase, thesis and blueprint documents from the end of the previous session. At any point I can initiate "Save State 1", 2, 3, etc. You don't want to load too many of these up without clearing them out along the way though, because it causes a lot of bloat on the server side. (I've been told this, but I've had as many as 4 or 5 save states created in a session before I started culling them).
I've only ever rolled back to a Save State once before, when things got really jacked up. The problem with the save states is that I don't want to allow it to have free access to create them whenever it feels like it or it does it all the time and if that bloat really is a problem it would happen almost right away. If I don't remember to make one fairly frequently though, I could end up losing a lot of good stuff we did if I were ever forced again to use one. I should probably add a rule to make the AI prompt me to make a save state every 2 hours or so, although with our new revision numbers on code changes it's probably not as important as it used to be.
What is cool is that it's pretty good at figuring out what we were trying to do with the codebase in previous versions.
I told you that this session had cut our codebase down to 33kb from a high of 56kb. That was true. But unfortunately a lot of things it thought was just "garbage" wasn't garbage, and we ended up adding most of it back. Along with the additions of around 5 feature requests this session, we've added back so much (in a different and more efficient way) that we're back up to 51kb now. But when adding back things that we had already figured out in our old spaghetti code, when provided with old code, the new session is able to say "oh, I see what you accomplished here. This is how we're going to do it now".
That's an annoying way to spend a day for sure, but there's no arguing that the codebase is much better for it. As I mentioned, the code will run about 5 times as fast as it used to and you don't hear that old hard drive chugging hard the whole time anymore. Everything works as it did before (and better than ever) now except for the tweaks we made to automatic processes that enforce certain thresholds for human operator review. Right now that code is still far too aggressive, and I need to get the guardrails back in now.
Oh... and I've had this session reorganize our historical documents. The blueprint now has an additional section that
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Quote:
Originally posted by SIGNYM:
I think our heat must have rolled your way, SIX. We're enjoying 85°F weather.
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
Uggh.... The high today is 97!!! On September 2nd!!!!
Gross.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Out soon on what has by the look of it a wet morning. Need some groceries and my area is sitting at 66F. Suppose to get rain today.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
All done for today and missed the rain by half an hour.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Mah jong in the morning in the rain.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Man...
I just got scammed by Scamazon.
Last night I realized that I was out a working coffee maker and needed to make an emergency purchase.
The last one I bought was the absolute WORST POS coffee maker I've ever had, which was an Amazon Basics branded unit on sale for less than $20. Four and a half star rating too! Whoooooo!
The thing had the shittiest pot with the thinnest pouring spout imaginable, so unless you spent a half a minute slowly pouring the thing into the back, you were liable to spill 10% of the water outside the unit. NUMBER ONE GRIPE ABOUT COFFEE POTS, IMO.
Then it wasn't long before the spring mechanism that allows you to pour a cup while the pot is still brewing just snapped right off. I was able to get it back in using some twist ties and it kind of worked, but not really.
And last night I noticed my coffee had gone cold while the light was still on, and later when I went to make another pot it was just dead. Nothing but the light worked.
So I sprang for a $30 Black and Decker this time.
I open it up today...
IT'S THE EXACT SAME GODDAMNED MACHINE.
I'm not joking. The ONLY difference is it's got the Black and Decker branding on the maker and the pot instead of Amazon Basics.
I'm glad that dude burned their warehouse down earlier this year.
Fuck Amazon.
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
It's all made in China, no matter whose name is on it.
We bought Agilent gas chromatographs and mass spectrometers for our lab. We could tell which ones were made in China bc the serial # began with a "C" (as opposed to a "U") and there was ALWAYS something effing wrong eith the China units! Either the flow controllers were defective or the internals were dirty with trace machining oil and needed two weeks of baking our before use.
You'd think companies like Agilent and Krups would enforce better QC tho!
-----------
"It may be dangerous to be America's enemy, but to be America's friend is fatal."- Henry Kissinger
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME
Turns out, even a good deal of the internals of our voting machines are made in China.
The machines that have internet connectivity.
Who knew?
--------------------------------------------------
Those who dance always seem crazy to those who can't hear the music.
NOTIFY: Y | REPLY | REPLY WITH QUOTE | PERMALINK | TOP | HOME