OpenAI researchers warned board of AI breakthrough ahead of CEO ouster, sources say (www.reuters.com)
from btp@kbin.social to technology@lemmy.ml on 23 Nov 2023 16:22
https://kbin.social/m/technology@lemmy.ml/t/650616

Ahead of OpenAI CEO Sam Altman’s four days in exile, several staff researchers wrote a letter to the board of directors warning of a powerful artificial intelligence discovery that they said could threaten humanity, two people familiar with the matter told Reuters.

After being contacted by Reuters, OpenAI, which declined to comment, acknowledged in an internal message to staffers a project called Q* and a letter to the board before the weekend’s events, one of the people said. An OpenAI spokesperson said that the message, sent by long-time executive Mira Murati, alerted staff to certain media stories without commenting on their accuracy.

Some at OpenAI believe Q* (pronounced Q-Star) could be a breakthrough in the startup’s search for what’s known as artificial general intelligence (AGI), one of the people told Reuters. OpenAI defines AGI as autonomous systems that surpass humans in most economically valuable tasks.

Given vast computing resources, the new model was able to solve certain mathematical problems, the person said on condition of anonymity because the individual was not authorized to speak on behalf of the company. Though only performing math on the level of grade-school students, acing such tests made researchers very optimistic about Q*’s future success, the source said.

Reuters could not independently verify the capabilities of Q* claimed by the researchers.

#technology

threaded - newest

autotldr@lemmings.world on 23 Nov 2023 16:25 next collapse

This is the best summary I could come up with:


Before his triumphant return late Tuesday, more than 700 employees had threatened to quit and join backer Microsoft (MSFT.O) in solidarity with their fired leader.

The sources cited the letter as one factor among a longer list of grievances by the board that led to Altman’s firing.

According to one of the sources, long-time executive Mira Murati told employees on Wednesday that a letter about the AI breakthrough called Q* (pronounced Q-Star), precipitated the board’s actions.

The maker of ChatGPT had made progress on Q*, which some internally believe could be a breakthrough in the startup’s search for superintelligence, also known as artificial general intelligence (AGI), one of the people told Reuters.

Given vast computing resources, the new model was able to solve certain mathematical problems, the person said on condition of anonymity because they were not authorized to speak on behalf of the company.

Though only performing math on the level of grade-school students, acing such tests made researchers very optimistic about Q*’s future success, the source said.


The original article contains 293 words, the summary contains 169 words. Saved 42%. I’m a bot and I’m open source!

ElectricAirship@lemmy.dbzer0.com on 23 Nov 2023 17:37 next collapse

What are they smoking at OpenAI? Can I get some?

malijaffri@feddit.ch on 23 Nov 2023 18:20 next collapse

Doesn’t WolframAlpha already do this?

NounsAndWords@lemmy.world on 23 Nov 2023 19:09 collapse

A calculator does most of it too, but this is a LLM that can do lots of other things also, which is a big piece of the “general” part of AGI.

Richard Feynman said “You have to keep a dozen of your favorite problems constantly present in your mind, although by and large they will lay in a dormant state. Every time you hear or read a new trick or a new result, test it against each of your twelve problems to see whether it helps. Every once in a while there will be a hit, and people will say, “How did he do it? He must be a genius!”

We are close to a point where a computer that can hold all the problems in its “head” can test all of them against all of the tricks. I don’t know what math problems that starts to solve but I bet a few of them would be applicable to cryptology.

But then again, I have no idea what I’m talking about and just making bold guesses based on close to no information.

p03locke@lemmy.dbzer0.com on 24 Nov 2023 16:33 next collapse

And he said this in the 80s, when AI as we know it today was barely a concept.

malijaffri@feddit.ch on 25 Nov 2023 19:14 collapse

Even so, I think I’ll hold off on calling anything AGI until it can at least solve simple calculus problems with a 90% success rate (reproducibly). I think that’s a fair criteria, in my opinion.

NounsAndWords@lemmy.world on 25 Nov 2023 19:21 collapse

I’d say more than that. I don’t think anyone is that close to AGI…yet

eestileib@sh.itjust.works on 23 Nov 2023 19:01 next collapse

Deep Mind is actually delivering shit like an estimate of the entire human proteome structure and creating the transcendently greatest go player of all time.

Meanwhile these chucklefucks are using the same electricity demand as Belgium to replicate a math solver that could probably be assigned as half-term project in an undergraduate class and are pissing themselves about threatening humanity.

The Valley has lost its goddamn mind.

332@feddit.nu on 23 Nov 2023 20:03 next collapse

Nicely put.

AngrilyEatingMuffins@kbin.social on 23 Nov 2023 21:10 next collapse

The computer self corrected based on its understanding of math principles that it learned through text. It’s not about the math. It used reason.

The computer had a thought. A rudimentary one, yes. But an actual thought.

I don’t really know what to say if you don’t see why that’s an amazing discovery.

Also the Belgium thing was if it continued growing at the rate it is, but the technology didn’t improve. The technology has already improved by two generations since that paper was written. It’s a crappy talking point and nothing else.

technojamin@lemmy.world on 23 Nov 2023 21:30 next collapse

While I agree that a lot of the hype around AI goes overboard, you should probably read this recent paper about AI classification: arxiv.org/abs/2311.02462

Systems like DeepMind are narrow AI, whereas LLMs are general AI.

spark947@lemm.ee on 24 Nov 2023 17:07 collapse

Not really. The imementation of land is most the same, they just run continously on a per word (token) basis.

sincle354@kbin.social on 24 Nov 2023 02:55 next collapse

Valley bullshit aside, I do have to defend the expensive exploration of the generalized AI space purely because it's embarassingly parallel. That is, it just gets so much better the more money and resources you throw at it. It couldn't solve math without a few million dollars worth of supercomputer training time. We didn't know it would create valid VHDL-to-csv-to-VBA scripts, but I got phind(.com) to make me one. And I certainly can't tell Wolfram Alpha to package the math solution it generated as a Javascript function.

Redex68@lemmy.world on 24 Nov 2023 12:38 next collapse

You are missing the very crucial part about how this is generalised. That’s like saying we don’t need to teach math to people anymore, we have calculators now. The AI isn’t too capable currently, but dismissing it would be like dismissing consumer PCs, because what are people gonna do with computers?

p03locke@lemmy.dbzer0.com on 24 Nov 2023 16:29 collapse

Deep Mind is actually delivering shit like an estimate of the entire human proteome structure and creating the transcendently greatest go player of all time.

Not to mention the huge advances in Chess AI. LeelaChessZero is the open-source implementation of the original AlphaZero idea Google came out with, and is rivaling Stockfish 15. Meanwhile, Torch is a new AI being developed that is now kicking Stockfish’s ass.

Grandmasters and notices alike are learning a lot from chess AI, figuring out better ways to improve themselves, either by playing them outright, using them for post-game analysis, or watching two bots play and see the kind of creative strategies they can come up with.

NotTheOnlyGamer@kbin.social on 24 Nov 2023 00:59 collapse

I think it's time to shut it down, hard. That's the start of something that will not end well for human beings.