That's been my assumption since the beginning of this drama last year. He seems to have one goal: real AGI. He knows that while LLMs may make something that seems like AGI there's nothing actually intelligent about it and its never going to get them there. OpenAI wants to pivot and sell sell sell because all they see is potential trillions of dollars, and it's time to make money instead of burning more millions/billions chasing a dream.

Yet all the AI weirdos on Twitter seem convinced that Ilya "saw something" (AGI) and got scared and wanted to pull the plug...lmao.

This is counter to every interview ilya has ever given since gpt3--he believes scaling llms can get there that's why they scaled to gpt4 scales at all.

There is more money in AGI than LLMs.

Whatever it is, language seems key to intelligent algorithms.

Nah, he departed due to politics (failed coup) and shift from research first to profit first. Same with Karpathy I believe.

He’ll most likely go somewhere where he can get a lot of compute and go back to research first.

> There is more money in AGI than LLMs.

I'd argue there's no money in AGI. Such a technology actually being available would be such a seismic shift in society that it breaks every commonly accepted economic norm we have today.

Real AGI means ALL knowledge workers are effectively out of a job. And AGI + advanced robotics means everyone is out of a job. Once no one has jobs, money loses all purpose.

Whatever company invented AGI, if it ever happens, may rule the world but it won't be with money.

> There is more money in AGI than LLMs.

That doesn't matter, they know how to do LLMs, they don't know how to do AGI. To modern capitalists that means you invest in one and the other someone else can do.

OpenAI was exploratory until they found something that could make them rich, then they went closed and for profit, now you should see them as just another for profit.

They are a nonprofit.

Like Google?

> it's time to make money instead of burning more millions/billions chasing a dream.

The investors want their money before people realise they have been oversold the dream/threat of AGI.

> Yet all the AI weirdos on Twitter seem convinced that Ilya "saw something" (AGI) and got scared and wanted to pull the plug...lmao.

The market to believe in made-up stories is alaways strong.

Isn't it consensus that AGI will never arise from LLMs?

Just based on current energy usage never going to happen. You just have to ask them to show you their energy bills alongside their demos.

A plane takes a LOT more energy than a bird, yet both fly, one of them a lot faster.

A plane is good at hauling cargo and goes fast, but general flying skills? Doesn't even come close.

Exactly, a giant LLM is not even close to be as power efficient as a human brain, in the same vein as a plane isn't as good at general flying. Yet it provides huge value and in many dimensions (that are important to us as humans) a plane can do way more than any bird.

Honestly, is ChatGPT providing huge value yet though ?

It’s cool and all. But I’m not sure if I’d say it’s as useful as flight yet.

If I could choose between LLm access and aviation, I’d stick with aviation.

There is no such consensus.

Great point. I should have phrased it "majority consensus", my bad.

No... A good number of folks will even go so far as to say that "all we do" is token prediction too. It's worth noting -- OpenAI founder Elon Musk claims in a lawsuit that the company has achieved AGI. Make of that what you will, but certainly there are many people on this site who believe in the general potential of LLMs.

Elon also claims that Tesla has solved full self driving, has announced robotaxis, while the SEC is investigating him for "inappropriate forward-looking statements".

Yeah, I study* that way too.

*procrastinate

> Yeah, I study* that way too

I’m not the only one?

So this is what it’s like when doves cry! - Milhouse

Funny