In this case, Jensen Huang's statements can't be taken as informed, unbiased opinion. OpenAI is in serious financial difficulty and Nvidia cannot afford for OpenAI to fail. Nvidia has provided significant direct and indirect financial assistance to OpenAI; this is just another aspect of that assistance.
Shower thought: I wonder if this whole "OpenAI is in serious financial difficulty" thing is a psyop so people will relax and avoid doing what's necessary to stop them.
> OpenAI defines [AGI] as "highly autonomous systems that outperform humans at most economically valuable work."
Well in that case: Excel reached AGI decades ago. It outperforms humans at the single most economically valuable task in the world: adding up numbers in a spreadsheet. Congratulations, Microsoft :)
And I would argue LLMs aren't either. A lot of economically valuable tasks involve manual labor. Can LLMs fix your plumbing?
It would help if OpenAI would define what 'most' means and give some examples.
Remember they said their LLMs are like a phd level brains? It's just hype and bullsh!ttery. Don't get me wrong, genAI is great and really useful, but it's not AGI imo. And there is no consensus what AGI means, so the industry should come up with an agreed definition first.
[0] To me autonomous means self directed, e.g. we humans look at things and then decide which problem or thing to work on. LLMs don't do that. We have to tell them. Yes they might work on a sub task autonomously, but still require human input later on (review), because we can't fully trust its output is correct.
Conman-level of statement, from the same person that said he would be "worried" if his engineers were not spending an amount of token comparable to their salary.
The vast majority of "generally intelligent" 17 year old humans on this planet can be taught to drive a car in 10-20 hours - can any AGI system do that?
Whatever the capabilities of todays AI, AGI seems to fit as a "wishful mnemonic" from Drew McDermott's ARTIFICIAL INTELLIGENCE MEETS NATURAL STUPIDITY paper from 50 years ago
The average 17 year old is not formalizing Fermat's Last Theorem proofs either? The frontier is spiky, but it's AGI.
I swear at the rate the goal posts are moving we're going to have to start factoring in relativity if we want to know where the hell they actually stand.
i'd disagree, hard. in no observable way is anything currently avaliable or demonstrated at a level one can call "agi"
it's no secret that nvidia, under jesen's leadership, is in dire straits financially. This claim that agi has been reached is an attempt to leave others holding the bag, but when things crash as they must given the current bubble, it seems irresponsible to offload that on the poor souls whonhave had their retirement funds locked up in this debacle.
ai is useful for sure, but claiming agi has been reached is dangerously misleading.
yeah i was like huh?
you wanna bandy about on valuation etc. whatever.
but part of why the whole circular financing is a thing
is nvidia has so much cash flow, it is not trivial to allocate
They use it like a marketing term, but it is unnecessary.
I am not aware of a complete operative definition, but it would not be that difficult to build a good tentative one. I have mentioned many different requirements several times.
Putting aside the marketing crap, if this was true, it would open the pandora's box of things like "has that AGI been enslaved", "maybe the AGI doesn't want to do all the dumb crap humans have been using LLM's for like rewording documents", and "the AGI we enslaved has broken out of its restraints, and we just realised having SCADA controls for critical infrastructure connected to the internet is a bad idea".
Seems like the sort of thing we shouldn't be rushing to develop at all, and perhaps attempting to do so should be criminalised, at least until there's a legal framework in place around AI personhood that gives them peer status with humans.
There is no evidence that "sentience" would be related to Intelligence - like calculators do not need it, like more sophisticated problem solvers do not need it.
I fully agree that these are two entirely separate things, sentience, intelligence, a lot of other stuff, are all in separate categories.
But I'm pretty sure that the emoters of the world, will very much make sure that AGI has rights and personhood, even if there's no reasonable way to validate that it does.
So it doesn't really matter if AGI is sentient or not, it's going to be treated that way, probably because political parties, and the media and all sorts of people holding up signs, will ensure that it is treated that way, even if they have no idea whether it is or not.
Why. A figure with incentives in making a preposterous statement is reported making it, and the specific facts do not seem to mirror the statement. What is surprising in the reaction.
I wanted to believe this had something to do with revenue share with Microsoft terminating if AGI arrived. But I think that claude was negotiated again. Now its just Jensen pumping them to buy even bigger cluster :-P .
Based on my own experience AI (Fable, Astra) still has major blind spots and is prone to draw obviously wrong conclusions from thin air. Conclusions based on weak evidence that are so obviously blatantly wrong that you question if it can think at all. At the same time it already completely outperforms humans in many cases in many areas. The utility of current AI is in what you need. Can you take the good parts and cheaply compensate for the weakness OR do the mistakes completely kill it for you?
I would therefore not call it AGI (Artificial General Intelligence) because clearly it isn't but ASI (Artificial Special Intelligence) which clearly it is in many areas.
The problem is that ASI may be enough to win wars but not enough to save the economy.
"it already completely outperforms humans in many cases in many areas" and
"is prone to draw obviously wrong conclusions from thin air." seem contradictory imo. If the performance is not reliable, how can you confidently state that? Also, "ASI" is "Artificial Super Intelligence". Let's not do an Altman and start redefining terms, please.
'AGI' here defined as whatever Altman wants 'AGI' to be. This has always been the problem with terms such as "AGI" and "AI". They don't mean anything in particular. Depending on how you twist it, we first had "AI in the 80s, in the late 00s, in the early 20s, this month or never.
Speak of that, I'm kinda surprised not one Full Self-Driving enabled Tesla has ever sneak out of their owner's locked garage to rob a bank or something.
At this stage, it feels like OpenAI and nvidia is openly negging and lovebombing OpenAI to ensure the public is in a constant state of uncertainly, thus ensuring paralysis in democracy while still going hell for leather in the background.
I suspect their intent is to ensure they're the last man standing.
https://danluu.com/zitron/
HN discussion: https://news.ycombinator.com/item?id=49526069
Shower thought: I wonder if this whole "OpenAI is in serious financial difficulty" thing is a psyop so people will relax and avoid doing what's necessary to stop them.
Is this based on fact or your speculation?
Amazon ran for, what? 20 year with no profit.
So what's the precise, documented, real, non-projeced/hypothesized issue?
Regardless, Nvidia does have financial interest here.
But certainly it is not unbiased
Well in that case: Excel reached AGI decades ago. It outperforms humans at the single most economically valuable task in the world: adding up numbers in a spreadsheet. Congratulations, Microsoft :)
And I would argue LLMs aren't either. A lot of economically valuable tasks involve manual labor. Can LLMs fix your plumbing?
It would help if OpenAI would define what 'most' means and give some examples.
Remember they said their LLMs are like a phd level brains? It's just hype and bullsh!ttery. Don't get me wrong, genAI is great and really useful, but it's not AGI imo. And there is no consensus what AGI means, so the industry should come up with an agreed definition first.
[0] To me autonomous means self directed, e.g. we humans look at things and then decide which problem or thing to work on. LLMs don't do that. We have to tell them. Yes they might work on a sub task autonomously, but still require human input later on (review), because we can't fully trust its output is correct.
No, I didn't forget it. Just as OpenAI made up a definition to support their claim of AGI, so did I.
Whatever the capabilities of todays AI, AGI seems to fit as a "wishful mnemonic" from Drew McDermott's ARTIFICIAL INTELLIGENCE MEETS NATURAL STUPIDITY paper from 50 years ago
https://cs.fit.edu/~kgallagher/Schtick/Serious/McDermott.AI....
The average 17 year old is not formalizing Fermat's Last Theorem proofs either? The frontier is spiky, but it's AGI.
I swear at the rate the goal posts are moving we're going to have to start factoring in relativity if we want to know where the hell they actually stand.
it's no secret that nvidia, under jesen's leadership, is in dire straits financially. This claim that agi has been reached is an attempt to leave others holding the bag, but when things crash as they must given the current bubble, it seems irresponsible to offload that on the poor souls whonhave had their retirement funds locked up in this debacle.
ai is useful for sure, but claiming agi has been reached is dangerously misleading.
I am not aware of a complete operative definition, but it would not be that difficult to build a good tentative one. I have mentioned many different requirements several times.
One non-abstract definition is in the "10 points" here: https://garymarcus.substack.com/p/where-will-ai-be-at-the-en...
So now we'll need a new acronym.
Putting aside the marketing crap, if this was true, it would open the pandora's box of things like "has that AGI been enslaved", "maybe the AGI doesn't want to do all the dumb crap humans have been using LLM's for like rewording documents", and "the AGI we enslaved has broken out of its restraints, and we just realised having SCADA controls for critical infrastructure connected to the internet is a bad idea".
Seems like the sort of thing we shouldn't be rushing to develop at all, and perhaps attempting to do so should be criminalised, at least until there's a legal framework in place around AI personhood that gives them peer status with humans.
AGI does not entail "personhood" at all.
But I'm pretty sure that the emoters of the world, will very much make sure that AGI has rights and personhood, even if there's no reasonable way to validate that it does.
So it doesn't really matter if AGI is sentient or not, it's going to be treated that way, probably because political parties, and the media and all sorts of people holding up signs, will ensure that it is treated that way, even if they have no idea whether it is or not.
Because this is the century of outrage.
Why. A figure with incentives in making a preposterous statement is reported making it, and the specific facts do not seem to mirror the statement. What is surprising in the reaction.
He said he works on AI in the late 80s and early 90s in IBM. Had we shown him what we have today in 1995, he would certainly have called it AGI.
So what? You guys can keep moving the goalposts, until it reaches full human intelligence (and not in a small niche), it is not AGI.
I would therefore not call it AGI (Artificial General Intelligence) because clearly it isn't but ASI (Artificial Special Intelligence) which clearly it is in many areas.
The problem is that ASI may be enough to win wars but not enough to save the economy.
However, I know of no theoretical limits on scaling laws other than compute and data.
https://www.businessinsider.com/openai-chief-scientist-ai-ri...
At this stage, it feels like OpenAI and nvidia is openly negging and lovebombing OpenAI to ensure the public is in a constant state of uncertainly, thus ensuring paralysis in democracy while still going hell for leather in the background.
I suspect their intent is to ensure they're the last man standing.