The argument for current LLM AIs leading to AGI has always been that they would spontaneously develop independent reasoning, through an unknown emergent property that would appear as they scale. It hasn’t happened, and there’s no sign that it will.
That’s a dilemma for the big AI companies. They are burning through billions of dollars every month, and will need further hundreds of billions to scale further - but for what in return?
Current LLMs can still do a lot. They’ve provided Level 4 self-driving, and seem to be leading to general-purpose robots capable of much useful work. But the headwinds look ominous for the global economy, - tit-for-tat protectionist trade wars, inflation, and a global oil shock due to war with Iran all loom on the horizon for 2025.
If current AI players are about to get wrecked, I doubt it’s the end for AI development. Perhaps it will switch to the areas that can actually make money - like Level 4 vehicles and robotics.
I don’t think anyone in the industry thought LLMs were going to reach AGI. But LLMs will be useful as part of an AGI framework. That’s the current focus in the industry.
I mean, yeah, this is about people outside the industry, those who invested money.
To my knowledge, LLMs still don’t pay for themselves. When the hype dies down and investors aren’t willing to provide money anymore, then prices for LLMs will become prohibitive for many current use-cases. That will also shrink the industry.
What that means in effect, we’ll still have to see, but AGI was one path that investors hoped for to get towards profitability, so it doesn’t aid the hype when they’re slowly learning about reality.
I’m developing some human centric LLM frameworks at work. Every API request to OpenAI is currently subsidized by venture capital. I do worry about what the industry will look like once there is a big price adjustment. Locally run models are pretty decent now and the pace is still moving forward, especially with regards to context window sizes so as long as I keep the frameworks model agnostic it might not be a big impact.
Yep, based off public numbers regarding OpenAI finances, it was estimated that as of a few months ago they spent $2.35 for every $1 they made.
It’s what Altman has constantly said was going to happen. Up to you to decide if he’s actually in the industry or not.
When has Sam Altman said LLMs will reach AGI? Can you provide a primary source?
He has said that they already know everything they need to know to get to AGI.
OpenAI has not made a single thing that wasn’t just a wrapped LLM.
So either A: he has somehow been running a skunk works that has fundamentally changed everything we know limits LLMs and none of the researchers leaked anything or B: he thinks LLMs are the way.
Additional quote when he was asked about AGI:
This tracks with the whole “I need 1 trillion dollars of energy investments” plan to get to AGI that he’s asked for.
The guy’s either honestly convinced himself that we can get there with deep learning scaling or he’s a conman. Could be both.
OpenAI has absolutely made non LLM products lmao
I’m not defending Sam Altman or the AI hype. A framework that uses an LLM isn’t an LLM and doesn’t have the same limitations. So the accurate media coverage that LLMs may have reached a plateau doesn’t mean we won’t see continued performance in frameworks that use LLMs. OpenAI’s o1 is an example. o1 isn’t an LLM, it’s a framework that augments some of the deficiencies of LLMs with other techniques. That’s why it doesn’t give you an immediate streamed response when you use it, it’s not just an LLM.
He said it again a few days ago on a Reddit AMA.
Perhaps the most interesting comment from Altman was about the future of AGI - artificial general intelligence. Seen by many as the ‘real’ AI, this is an artificial intelligence model that could rival or even exceed human intelligence. Altman has previously declared that we could have AGI within “a few thousand days”.
When asked by a Reddit user whether AGI is achievable with known hardware or it will take something entirely different, Altman replied: “We believe it is achievable with current hardware.”
https://www.techradar.com/computing/artificial-intelligence/chatgpt-5-wont-be-coming-in-2025-according-to-sam-altman-but-superintelligence-is-achievable-with-todays-hardware
You’ve completely misunderstood as others have pointed out to you. It’s great that you want to learn about this stuff, but you have a long ways to go before you are at a point to talk authoritatively about it. The best thing for you to do now is to set aside all your preconceived notions and start from the beginning with an open mind. There is no point in talking authoritatively before you spend some time learning.
That’s not Sam Altman saying that LLMs will achieve AGI. LLMs are large language models, OpenAI is continuing to develop LLMs (like GPT-4o) but they’re also working on frameworks that use LLMs (like o1). Those frameworks may achieve AGI but not the LLMs themselves. And this is a very important distinction because LLMs are reaching performance parity so we are likely reaching a plateau for LLMs given the existing training data and techniques. There is still optimizations for LLMs like increasing context window sizes etc.
It’s remarkable that anyone would think sam said that or thought that. It’s like there is a whole other universe where 3rd and 4th hand sources are treated like 1st hand.
No they did for sure. Ofcourse that was conditional on a few things. Those things have yet to arrive and one major detractor is actually the lack of training data. Most of the public internet has been crawled by AI crawlers and half the new content is poisoned by AI making it worthless. Its gonna take a few more years to see if it keeps scaling or not.