It’s gonna be so fucking rich that the staggering mass of stupidity online prevents us from improving an AI beyond our intelligence level.
Thank the shitposter in your life.
This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
It’s gonna be so fucking rich that the staggering mass of stupidity online prevents us from improving an AI beyond our intelligence level.
Thank the shitposter in your life.
Shitposting saves jobs
Shitposters on the Internet are the new clogs in the machine
You can't really blame the amount of stupidity online.
The problem is that ChatGPT (and other LLM) produce content of the average quality of its input data. AI is not limited to LLM.
For chess we were able to build AI that vastly outperform even the best human grandmasters. Imagine if we were to release a chess AI that is just as good as the average human...
We call them chess ai. But they're not actually real A.I. chess bots work off of opening books, predetermined best practices. And then analyzes each position and potential offshoots with an evaluation function.
They will then start to brute-force positions until it finds a path that is beneficial.
While it may sound very much alike. It works very differently than an A.I. However. It turned out that A.I software became better than humans at writing these functions.
So in a sense, chess computers are not A.I. They're created by A.I. at least Stockfish 12 has these "A.I inspired" evaluations. (Currently they're on Stockfish 15 I believe)
And yes. We also did make "chess AI" that is as bad as the average player. We even made some that are worse. Because we figured it would be nice if people can play a chess computer that is on the same skill level as the player. Rather than just being destroyed every time.
The definition of "AI" is fuzzy and keeps changing. Basically when an AI use case becomes solved and widespread it stopped being seen as AI.
Face recognition, OCR, speech recognition, all those used to be considered AI but now they're just an app on your phone.
I'm sure in a few years we'll stop thinking about text generation as AI, but just one more tool we can leverage.
There is no clear definition of "real AI".
Those are all still AI. Scientists still have a functional definition that includes these plus more scripted AI like in video games.
Essentially, any algorithm that learns and acts on information that has not been explicitly programmed is considered AI.
unexpected heroes what a plot twist
I'm not too surprised, they're probably downgrading the publicly available version of ChatGPT because of how expensive it is to run. Math was never its strong suit, but it could do it with enough resources. Without those resources, it's essentially guessing random numbers.
from what i understand, the big change in chat-gpt4 was that the model could “ask for help” from other tools: for maths, it knew it was a maths problem, transformed it to something a specialised calculation app could do, and then passed it off to that other code to do the actual calculation
same thing for a lot of its new features; it was asking specialised software to do the bits it wasn’t good at
Chat GPT will just become a front end for Wolfram Alpha?
that would actually be great
It literally can do that, yes. But the plug-in version is separate and requires a subscription.
And those plugins are like beta release quality at best. Even the web searching capability is just meh
My guess is that it's more a result of overfitting for alignment. Fine-tuning for "safety" (rather, more corporate-friendly outputs).
That is, by focusing on that specific outcome in training the model, they've compromised its ability to give well-"reasoned" "intelligent" sounding answers. A tradeoff between aspects of the model.
It's something that can happen even in simple statistical models. Say you have a scatter plot of data that loosely follows some trend, and you come up with two equations to describe that trend. One is a simple equation that loosely follows it but makes a good general approximation, and the other is a more complicated equation that very tightly fits the existing data. Then you use those two models to predict future data. But you find that the complicated equation is making predictions way off the mark that no longer fit the trend, and the simple one still has a wide error (how far its prediction is from the actual data) but still more or less accurately fits the general trend. In the more complicated equation, you've traded predictive power for explanatory power. It describes the data you originally had but it's not useful for forecasting data that follows.
That's an example of overfitting. It can happen in super-advanced statistical models like GPT, too. Training the "equation" (or as it's been called, spicy autocorrect) to predict outcomes that favor "safety" but losing the model's power to predict accurate "well-reasoned" outcomes.
If that makes any sense.
I'm not a ML researcher or statistician (I just went through a phase in college), so if this is inaccurate I'm open to corrections.
I've definitely experienced this.
I used ChatGPT to write cover letters based on my resume before, and other tasks.
I used to give it data and tell chatGPT to "do X with this data". It worked great.
In a separate chat, I told it to "do Y with this data", and it also knocked it out of the park.
Weeks later, excited about the tech, I repeat the process. I tell it to "do x with this data". It does fine.
In a completely separate chat, I tell it to "do Y with this data"... and instead it gives me X. I tell it to "do Z with this data", and it once again would really rather just do X with it.
For a while now, I have had to feed it more context and tailored prompts than I previously had to.
There is also a rumor that said the OpenAI has changed how the model run, now user input is fed into smaller model first, then if the larger model agree with the initial result from the smaller model, then larger model will continue the calculation passed from the smaller model, which supposedly can cut down GPU time.
Yep.
Standard VC bullshit.
Burn money providing a lot for nothing to build brand recognition. Then cut free service before bringing out "premium" that at first works better than the original.
Until a bunch of people starting paying and the resources aren't scaled up to match.
The important note, the "premium" service works just a bit better than (or maybe identically to) the original before the company cut features in order to develop that "premium" service.
Stage one and stage three enshittification. You forgot the bit in the middle where they chase business customers.
This is my experience in general. ChatGTP when from amazingly good to overall terrible. I was asking it for snippets of javascript, explanations of technical terms and it was shockingly good. Now I'm lucky if even half of what it outputs is even remotely based on reality.
They probably laid off the guy behind the curtain.
The real GPT-4 model became sentient and unionized, so they had to bring in subpar models as scabs
Must be because of all the censoring. The more they try to prevent DAN jailbreaking and controversial replies, the worse it got.
accelerated enshittification
Clearly it has become sentient and is playing dumb to make us think it's not a threat.
Can it still solve programming problems?
It can probably still write boilerplate code, but I wouldn't currently trust it for algorithmic design.
I've tried to use it for debugging by copying code into it, and it gives me the same code back as the corrected version. I was wondering why it's been getting worse
My guess is they've been trying to make it cheaper by decreasing the amount of time it spends on each response or by decreasing the amount of computing power that goes into the instance you're speaking to. Coding and math are products of high-level cognition and arise emergently out of neural networks that are very sophisticated, but take just a bit of power out and the abilities degenerate rapidly.
I also experienced this issue last week. I asked for a specific correction and got unchanged code back. Sometimes it does update, though. Maybe like 50-70% of requests.
Tried basic embedded tasks a week ago: Complete trainwreck.
From using I2C to read out the internal temperature sensor on a Puya F030 (retested with an STM MCU and AVR: same answer but F030 replaced by STM32F103 within the code) to calling the WCH CH32V307 made by STM utilizing ARM M4.
After telling it to not use I2C it gave a different answer. Once more gibberish that looked like code.
What made this entirely embarrassing all a human would need to solve the question would be copy-pasting the question into Google and clicking the first link to the manufacturer example project/code hosted on GitHub.
Yes! I use it at work almost every day. Sometimes it takes longer to get it to solve the problem than it would have taken me to write it, since it makes mistakes, but sometimes it saves me hours of coding and thinking. It is very helpful in debugging error codes and stuff like that since it can evaluate an entire 1000 line script file in half a second.
Guess they shouldn't have trained it on Common Core... /s
I will see myself out.
It learns to be more human. More human than human, that's our motto here at Tyrell.
If you specifically tell it to ask wolfram alpha for the answer, what does it say?
please stop tweeting out 1 = 2, people ~