- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
For OpenAI, o1 represents a step toward its broader goal of human-like artificial intelligence. More practically, it does a better job at writing code and solving multistep problems than previous models. But it’s also more expensive and slower to use than GPT-4o. OpenAI is calling this release of o1 a “preview” to emphasize how nascent it is.
The training behind o1 is fundamentally different from its predecessors, OpenAI’s research lead, Jerry Tworek, tells me, though the company is being vague about the exact details. He says o1 “has been trained using a completely new optimization algorithm and a new training dataset specifically tailored for it.”
OpenAI taught previous GPT models to mimic patterns from its training data. With o1, it trained the model to solve problems on its own using a technique known as reinforcement learning, which teaches the system through rewards and penalties. It then uses a “chain of thought” to process queries, similarly to how humans process problems by going through them step-by-step.
At the same time, o1 is not as capable as GPT-4o in a lot of areas. It doesn’t do as well on factual knowledge about the world. It also doesn’t have the ability to browse the web or process files and images. Still, the company believes it represents a brand-new class of capabilities. It was named o1 to indicate “resetting the counter back to 1.”
I think this is the most important part (emphasis mine):
As a result of this new training methodology, OpenAI says the model should be more accurate. “We have noticed that this model hallucinates less,” Tworek says. But the problem still persists. “We can’t say we solved hallucinations.”
You must log in or register to comment.
Interesting, thanks for sharing.
Gave it a try just now. Pretty terrible result.
Show it.
At the same time, o1 is not as capable as GPT-4o in a lot of areas. It doesn’t do as well on factual knowledge about the world. It also doesn’t have the ability to browse the web or process files and images. Still, the company believes it represents a brand-new class of capabilities. It was named o1 to indicate “resetting the counter back to 1.”
I think it’s more of a proof of concept then a fully functioning model at this point.
To be fair, I did ask it to fact check an article, so that was probably not the best first choice
Facts. A “reasoning AI” has problems with … lemme check this again … facts?
Find the comment about psychics, it’s exactly the situation we are currently in.
Reasoning has nothing to do with knowledge though.
So for those not familar with machine learning, which was the practical business use case for “AI” before LLMs took the world by storm, that is what they are describing as reinforcement learning. Both are valid terms for it.
It’s how you can make an AI that plays Mario Kart. You establish goals that grant points, stuff to avoid that loses points, and what actions it can take each “step”. Then you give it the first frame of a Mario Kart race, have it try literally every input it can put in that frame, then evaluate the change in points that results. You branch out from that collection of “frame 2s” and do the same thing again and again, checking more and more possible future states.
At some point you use certain rules to eliminate certain branches on this tree of potential future states, like discarding branches where it’s driving backwards. That way you can start opptimizing towards the options at any given time that get the most points im the end. Keep the amount of options being evaluated to an amount you can push through your hardware.
Eventually you try enough things enough times that you can pretty consistently use the data you gathered to make the best choice on any given frame.
The jank comes from how the points are configured. Like AI for a delivery robot could prioritize jumping off balconies if it prioritizes speed over self preservation.
Some of these pitfalls are easy to create rules around for training. Others are far more subtle and difficult to work around.
Some people in the video game TAS community (custom building a frame by frame list of the inputs needed to beat a game as fast as possible, human limits be damned) are already using this in limited capacities to automate testing approaches to particularly challenging sections of gameplay.
So it ends up coming down to complexity. Making an AI to play Pacman is relatively simple. There are only 4 options every step, the direction the joystick is held. So you have 4n states to keep track of, where n is the number of steps forward you want to look.
Trying to do that with language, and arguing that you can get reliable results with any kind of consistency, is blowing smoke. They can’t even clearly state what outcomes they are optimizing for with their “reward” function. God only knows what edge cases they’ve overlooked.
My complete out of my ass guess is that they did some analysis on response to previous gpt output, tried to distinguish between positive and negative responses (or at least distinguish against responses indicating that it was incorrect). They then used that as some sort of positive/negative points heuristic.
People have been speculating for a while that you could do that, crank up the “randomness”, have it generate multiple responses behind the scenes and then pit those “pre-responses” against each other and use that criteria to choose the best option of the “pre-responses”. They could even A/B test the responses over multiple users, and use the user responses as further “positive/negative points” reinforcement to feed back into it in a giant loop.
Again, completely pulled from my ass. Take with a boulder of salt.
Again, completely pulled from my ass. Take with a boulder of salt.
You’re under arrest. That’s ass-salt.
Fuck you, that made me smile. And I haven’t even had my coffee yet.
Sorry for hitting you at a vulnerable time.
To be a little nitpicky most of the AI that can play Mario kart are trained not with a reinforcement learning algorithm, but woth a genetic algorithm, which is a sort of different thing.
Reinforcement learning is rather like how you teach a child. Show them a bunch of good stuff, and show them a bunch of bad stuff, and tell them which is the good stuff and which is the bad stuff.
Genetic algorithms are where you just leave it alone, simulate the evolutionary process on an accelerated time scale, and let normal evolutionary processes take over. Much easier, and less processor intensive, plus you don’t need huge corpuses of data. But it takes ages, and it also sometimes results in weird behaviors because evolution finds a solution you never thought of, or it finds a solution to a different problem to the one you were trying to get it to find a solution to.
… sometimes results in weird behaviors because evolution finds a solution you never thought of, or it finds a solution to a different problem to the one you were trying to get it to find a solution to.
Those outcomes seem especially beneficial.
But it takes ages, …
Is this process something that distributed computing could be leveraged for, akin to SETI@home?
I work in computer science but not really anything to do with AI so I’m only adjacently knowledgeable about it. But my understanding is unfortunately, no not really. The problem would be that if you run a bunch of evolutions in parallel you just get a bunch of independent AIs, all with slightly different parameters but they’re incapable of working together because they weren’t evolved to work together, they were evolved independently.
In theory you could come up with some kind of file format that allowed for the transfer of AI between each cluster, but you’d probably spend as much time transferring AI as you saved by having multiple iterations run at the same time. It’s n^n problem, where n is the number of AIs you have.
Genetic algorithms is a sort of broad category and there’s certainly ways you could federate and parallelize. I think autoML basically applies this within the ML space (multiple trainings explore a solution topology and convergence progress is compared between epochs, with low performers dropping out). Keep in mind, you can also use a genetic algorithm to learn how to explore an old fashioned state tree.
I’m getting so tired of the pessimists who are against AI. Granted, I can reflect and see my own similar attitude towards Trump: no matter what, I would never vote for him considering his history and who he is as a person. But treating the next generation of technology feels different than that to me; AI is the future, it’s the next revolution. Sure, there are several real issues to criticize and question (copyright, compensation, hallucination come to mind) but instead shit here on Lemmy just gets downvoted to hell with no explanation. I know this comment will get downvoted, but I just wish we could have a discussion about the future without shutting down every practical comment wanting to talk about it.
More and more advanced tools for automation are an important part of creating a post-scarcity future. If we can combine that with tearing down our current economic system - which inherently requires and thus has to manufacture scarcity - we can uplift our species in ways we can currently only imagine.
But this ain’t it bud. If I ask you for water and you hand me a glass of warm piss, I’m not “against drinking water” for refusing to gulp it down.
This isn’t AI. It isn’t - meaningfully and usefully - any form of automation at all. A bunch of conmen slapped the letters “AI” on the side of their bottle of piss and you’re drinking it down like it’s grandma’s peach tea.
The people calling out the fundamental flaws with these products aren’t doing so because we hate the entire concept of automation, any more than someone exposing a snake-oil salesman hates medicine. What we hate is being lied to. The current state of this technology is bullshit and hype. It is not fit for human consumption (other than recreationally) and the money being pumped into it could be put to far better uses. OpenAI may have lofty goals, but they have utterly failed at achieving them, and right now any true desire to create AGI has been totally subsumed by the need to keep pumping out slightly better looking versions of the same polished turd in order to convince investors to keep paying for their staggeringly high hosting costs.
I’m kinda in the same boat but on the other side. I always try to argue with people about this. It gets me a lot of flak on pro AI posts but that won’t stop me. I usually get very aggressive replies and sometimes some fucked up dm’s too.
I’m against it because we are already seeing the consequences of this technology and it’s only getting worse. By the time laws catch up it’s gonna be too late and the damage will be done. For some technologies that’s not always the worst. But we already saw how long it took for anyone to do anything about the Internet when it came out, and we are still trying to this day. This shit is growing so fast we will all feel the whiplash. Sites like Facebook are getting absolutely flooded with so much AI that they are becoming almost unusable. And that’s before we even get into the shady shit people use AI for like making porn of people they know with the click of a button. I recently read an article about how bad deepfake porn is in South Korea (found the article. https://www.nytimes.com/2024/09/12/world/asia/south-korea-deepfake-videos.html). And in places like the US, where a lot of these companies are based, they are so slow to do anything about a problem it’s going to be too late by the time they get to it.
But besides all the awful things happening because of AI, I do have one personal gripe with the whole ordeal. Why are we so quick to replace the things we enjoy with AI? When I get home from work I like to make music and practice pixel art (I’m not very good at either yet). I’d much rather have AI replace my job than my hobbies. I’m down for things that are useful, but too much of this just gives me a bad gut feeling. Like their trying to replace people and not their jobs.
This may be the future. But it sounds like a pretty dystopian future to me. You already can’t believe everything you see on the Internet and this will only make it worse.
Being against LLMs being sold as AI (or as useful for anything practical) is not being against AI.
LLMs have absolutely nothing whatsoever to do with AI, other than being sold as if they did.
That’s not what reasoning is. Training is understanding what they’re talking about and being able to draw logical conclusions based on what they’ve learned. It’s being able to say, I didn’t know but wait a second and I’ll look it up," and then summing that info up in original language.
All Open AI did was make it less stupid and slap a new coat of paint on it, hoping nobody asks too many questions.
And this is something data scientists have already been doing with existing LLMs.
Terrence Tao shared his thoughs on Mastodon: https://mathstodon.xyz/@tao/113132502735585408
trained to answer more complex questions, faster than a human can.
I can answer math questions really really fast. Not correct though, but like REALLY fast!
I’m the same with any programming question as long as the answer is Hello World
That’s a flat out lie, I use it for code all the time and it’s fantastic at writing useful functions if you tell it what you want. It’s also fantastic if you ask it to explain code or options for problem solving.
😋
It scores 83% on a qualifying exam for the international mathematics olympiad compared to the previous model’s 13% so…
When you say previous model, you mean gemini with alpha geometry (an actual RL method)? Which scored a silver?
I mean not only google did it before, they also released their details unlike openai’s “just trust me bro, its RL”.
Openai also said that we should reserve 25k tokens for this “reasoning” and they will be charged the same as output tokens which is exorbitantly high (60$ for 1m tokens).
And the cherry on top is that they won’t even give us these “reasoning” tokens. How the hell am I supposed to improve my prompts if I can’t even see it? How would I reduce the hallucinations without it?
My personal experience is that, it does have an extra reasoning thing going for itself but in no way does it make openai’s tactics tolerable. The quality does not increase enough to justify its cost per token, let alone their “reasoning tokens” BS.
“We have noticed that this model hallucinates less,” Tworek says. But the problem still persists. “We can’t say we solved hallucinations.”
On one hand, yeah, AI hallucinations.
On the other hand, have you met people?
I’m hallucinating right now, WEEEeee…
Can’t wait to read about it telling someone to put glue on pizza.
This is smarter. Will tell you how to pump a Calzone full of glue.
And provide the logical reasoning behind it!
“This meal will stick to your ribs! No, really…”
I just love how people seem to want to avoid using the word lie.
It’s either misinformation, or alternative facts, or hallucinations.
Granted, a lie does tend to have intent behind it, so with ChatGPT, it’s probably better to say falsehood, instead. But either way, it’s not fact, it’s not truth, and people, especially schools, should stop using it as a credible source.
Being wrong is not the same as lying. When LLMs start giving wrong answers on purpose to mislead people we would have a big problem.
The thought of a maliciously deceptive AGI is terrifying to me. Many, many people will trust it until it’s too late.
There was a recent paper that argues ‘bullshitting’ is the most apt analogy. I.e. telling something to satisfy the other person without caring about the truth content of what you say
What about thw term “incorrect facts”?
I think I’ve used it if this is the latest available, and it’s terrible. It keeps feeding me wrong information, and when you correct it, it says you’re right… But if you ask it again, it again feeds you the wrong information.
if you ask it again, it again feeds you the wrong information
Well, it’s a LLM, they can’t learn anything without rebuilding the whole model from scratch, which I wouldn’t exactly call learning anyway… all they “know” is what word is most likely to follow a certain sequence of words according to their model.
Any other facts or information are completely inconsequential for their operation and results.
How much more time until they use the word “sentient”?
Is that even the goal? Do we want an AI that’s self aware because I thought that basically the whole point was to have an intelligence without a mind.
We don’t really want sapient AI because if we do that then we have to feel bad about putting it in robots and making them do boring jobs. Don’t we basically want guildless servants, isn’t that the point?
For the servants bots, yes no sentience. For my in house AI assistant robot buddy/butler/nanny/driver - also yes no sentience.
Yeah I was thinking more about it as marketing, than a real thing
deleted by creator
What we want doesn’t have any impact on what our corporate overlords decide to inflict on us.
They don’t want sapient AI either, why would they?
No one is trying for a self-aware artificial intelligence.
Not until it has senses, which it currently does not have.
I’m more concerned about them using the word “sapient.” My dog is sentient; it’s not a high bar to clear.
The meaning is ok. But “sentient” is so hot right now
Until the bubble bursts
I’d recommend everyone saying “it can’t understand anything and can’t think” to look at this example:
https://x.com/flowersslop/status/1834349905692824017
Try to solve it after seeing only the first image before you open the second and see o1’s response.
Let me know if you got it before seeing the actual answer.
This example doesn’t prove what you think it does. It shows pattern detection - something computers are inherently very well suited for - but it doesn’t demonstrate “reasoning” in any meaningful way.
You should really look at the full CoT traces on the demos.
I think you think you know more than you actually know.
Got a link to that?
Yep:
https://openai.com/index/learning-to-reason-with-llms/
First interactive section. Make sure to click “show chain of thought.”
The cipher one is particularly interesting, as it’s intentionally difficult for the model.
The tokenizer is famously bad at two letter counts, which is why previous models can’t count the number of rs in strawberry.
So the cipher depends on two letter pairs, and you can see how it screws up the tokenization around the xx at the end of the last word, and gradually corrects course.
Will help clarify how it’s going about solving something like the example I posted earlier behind the scenes.
Actually, they are hiding the full CoT sequence outside of the demos.
What you are seeing there is a summary, but because the actual process is hidden it’s not possible to see what actually transpired.
People are very not happy about this aspect of the situation.
It also means that model context (which in research has been shown to be much more influential than previously thought) is now in part hidden with exclusive access and control by OAI.
There’s a lot of things to be focused on in that image, and “hur dur the stochastic model can’t count letters in this cherry picked example” is the least among them.
I think if you can actually define reasoning, your comments (and those like yours) would be much more convincing. I’m just calling yours out because I’ve seen you up and down in this thread repeating it, but it’s a general observed of the vocal critics of the technology overall. Neither intelligence nor reasons (likewise understanding and knowing, for that matter) are easily defined in a way that is more useful than invoking spirits and ghosts. In this case, detecting patterns certainly seems a critical component of what we would consider to be reasoning. I don’t think it’s sufficient, buy it is absolutely necessary.
While truly defining pretty much any aspect of human intelligence is functionally impossible with our current understanding of the mind, we can create some very usable “good enough” working definitions for these purposes.
At a basic level, “reasoning” would be the act of drawing logical conclusions from available data. And that’s not what these models do. They mimic reasoning, by mimicking human communication. Humans communicate (and developed a lot of specialized language with which to communicate) the process by which we reason, and so LLMs can basically replicate the appearance of reasoning by replicating the language around it.
The way you can tell that they’re not actually reasoning is simple; their conclusions often bear no actual connection to the facts. There’s an example I linked elsewhere where the new model is asked to list states with W in their name. It does a bunch of preamble where it spells out very clearly what the requirements and process are; assemble a list of all states, then check each name for the presence of the letter W.
And then it includes North Dakota, South Dakota, North Carolina and South Carolina in the list.
Any human being capable of reasoning would absolutely understand that that was wrong, if they were taking the time to carefully and systematically work through the problem in that way. The AI does not, because all this apparent “thinking” is a smoke show. They’re machines built to give the appearance of intelligence, nothing more.
When real AGI, or even something approaching it, actually becomes a thing, I will be extremely excited. But this is just snake oil being sold as medicine. You’re not required to buy into their bullshit just to prove you’re not a technophobe.
Dang, OpenAI just pulled an Apple. Do something other people have already done with the same results (but importantly before they made a big fuss about it), claim it’s their innovation, give it a bloated name so people imagine it’s more than it is and produce a graph comparing themselves to themselves, hoping nobody will look at the competition.
Just like Apple, they have their own selling point, but instead they seem to prefer making up stuff while forgetting why people use em.
On a side note they also pulled an Elon. Where’s my AI companion that can comment on video in realtime and sing to me??? Ya had it “working” “live” a couple months ago, WHERE IS IT?!?
Pulled an Apple?
I know you hate apple because android is way better but people loved their ipods, iphones, airpods and apple watches. Sure those things were made before but Apple did make them better. So I don’t know what your point is.
Assuming I’m an android fan for pointing out that Apple does shady PR. I literally mention that Apple devices have their selling point. And it isn’t UNMATCHED PERFORMANCE or CUTTING EDGE TECHNOLOGY as their adds seems to suggest. It’s a polished experience and beautiful presentation; that is unmatched. Unlike the hot mess that is android. Android also has its selling points, but this reply is already getting long. Just wanted to point out your pettiness and unwillingness to read more than a sentence.
Meanwhile a bald turtle and his AI anime daughter on twitch can do exactly this, and he’s building her at home on nvidia GPUs.
(Vedal987 and Neuro-sama, if you’re curious)
Reinforcement learning my beloved ❤️
