Hacker News Reader

Comment by whynotminot | Hacker News Reader

These product-focused moves feel like a tacit admission that AGI is further away than they’ve been preaching.

If it was close at hand, spending precious resources on anything other than pursuing AGI wouldn’t make sense.

Does anyone seriously think we're anywhere close to AGI with LLMs? I know CEOs like to say things to blow smoke up investors asses, but does anyone with actual credibility think that?

Anecdata but I talk to lots and lots of AI/ML/DS engineers. Everyone knows the current LLM architecture won't work for AGI. All the "reasoning" models are just pseudoreasoning and there is severe data leakage and benchmaxxing when the companies tout the capabilities.

People really avoid considering what the word "general" implies. Yesterday I tried sending o3 a screenshot of some sheet music, asking for a midi file of how it sounds. Complete failure x3. Could not even get the value of the first note right. This is not "general" intelligence.

These models are notably terrible at music in every dimension.

Music is essentially mathematical. Weakness in math is being addressed by dedicated capabilities that are triggered by mathematical language in prompts, but because these models are actually terrible at math there is no lateral transfer of skill to the domain of music. That's my theory anyway.

I'm general intelligence, and could not in a million years do that.

I think actually you could do that if you wanted to; look up what notes mean, write some little program to make a sound if you had to. You could do it in a week if it was your only job.

but OP didn't give the AI a week it gave it less than a minute

Don't sell yourself short. If you have eyes and a voice, you can learn to read sheet music and hum a tune.

If you can't learn to read sheet music in a million years then you are not generally intelligent.

Every time you deny yourself this kind of agency you are denying yourself your humanity.

you wouldn't need a million years to learn to read sheet music and write midi

I believe Yann LeCun more than I do any CEO. And he has been pretty clear you’re not going to get AGI any time soon.

Previously discussed on HN:

https://news.ycombinator.com/item?id=39059897

Fifteen years ago I worked with a guy who, in retrospect, was very similar to an LLM. He was extremely verbally gifted and a vacuum cleaner for information. He could speak brilliantly about any topic he had been exposed to. He was a great person to send to a meeting, because he was great at answering questions coherently with the information he had on hand, and he always managed to make your ideas sound smarter than you could yourself. Based on that, you might think he sounds like a gifted human, until I tell you about his major weakness: if you asked him about something he didn't know about, he would often speak just as surely, fluently, and compellingly about it. He hallucinated just like an LLM, and that's why he was stuck in roles without a high level of responsibility despite his verbal gifts.

He was neither arrogant nor self-conscious. He treated his hallucinations as if they were the kinds of simple mistakes other people made, like, oops, I thought I understood this but I don't, no different from oops, I forgot my umbrella.

I sometimes wondered if he had a specific condition that made him the way he was, but I never doubted that he was human, with "general intelligence."

If you’ve spent any time around little kids, you’ve certainly seen that making shit up is a natural inclination of the human brain.

Ideally, as one’s intellect matures, one learns to stop doing that, and build coherent reasoning, only speak up when you know what you’re talking about.

Well, ideally. Many people never get to that stage.

I see a lot of replies suggesting agreement: LLMs are nowhere close to AGI.

I agree — it may well be a completely different path we need to go down to get to AGI ... not just throwing more resources at the path we've pioneered. As though a moon landing were going to follow Montgolfier's early balloon flights in "about five years".

At the same time, there is suddenly so much attention + money on AI that maybe someone will forge that new path?

"Money is All You Need".

Unfortunately the money is all chasing LLMs, not other AI approaches. Anyone with a different idea is frozen out, at least for a while. Whenever LLM disillusion finally sets in with the investor class, the question is whether other “AI” will be able to distinguish itself or if the money will just all dry up for another few decades.

No. People hype it but it's obvious we're hitting a wall with LLMs.

That being said, the "apps" that use LLMs coming out now are good. Not AGI good, but they do things, will be disruptive and have value.

And the money coming it could lead to new techniques and eventual AI. For now though, it looks like AI is transitioning into products and figuring out how to lower inference costs.

I do think LLMs will make incredible progress and we'll see lots of breakthroughs from it, but I agree it's nowhere close to AGI.

I'm not sure that matters though—if a technology can give humans what they want exactly when they want it, it doesn't matter if AGI, LLMs, humans, or some other technology is behind that.

Tyler Cowen claims it's already here. I don't agree with him but he has credibility with a lot of people.

depends on what we mean when define AGI

i think there's ample evidence to suggest that we're growing closer (3-5 year timeline?) to replacement-level knowledge workers in targeted fields with limited scope. i don't know that i would call that AGI? but i think it's fair to call it close.

thing is that has value, but compute ain't cheap and the value prop there is more of reducing payroll rather than necessarily scaling business ops. this move to me looks like a recognition that generalized AI on it's own isn't a force multiplier as long as you have bottlenecks that make it too pricey to scale activity by an order of magnitude or more.

I expect a fair number of non-technical LLM proponents, and probably some engineers as well, have likely built machines quite capable of helping them fool themselves that it is.

Nobody fully understands how human intelligence works. It's implausible we'll be able to replicate it or even come up with something better in the short term.

“If the brain were so simple we could understand it, we’d be so simple we couldn’t.”

Neuroscience is like a 150 years or so since the discovery of the neuron?

It'll take some time but we'll get there. Just not as soon as the AI hype will make you believe.

to answer that you have to define AGI first

That's not a prerequisite to refute a position, no.

The ball is in the other court - if one is working on AGI, it behooves one to know what one is aiming at (and I'd stake a fair wager that OpenAI et al have at this moment very little better picture of what AGI looks like than you or I)

defining what you mean is absolutely a prerequisite to someone else being able to figure out if you are right

To some extent, yes, but the onus here is on the party making wildly unverifiable claims, not the party refuting them

I remain unconvinced they're (the whole LLM/"Attention Is All You Need" industry) even barking up the right tree to build anything usefully-close to "AGI".

The idea that any situation or sensory input can be broken down into a sequence of tokens, and that action choice can be characterized by predicting a subsequent sequence of tokens in the same space, may well bear fruit.

But I think that a lot of people also buy into the idea that "text and image data from the web, and from historical chats, is the right/only way to generate the data set required," and it's a dangerous trap to fall into.

What do you think the correct, or at least a more likely tree is?

its certainly currently useful and also generally intelligent, right?

It's definitely useful.

It can answer specialized PhD level questions correctly, yet cannot perform tasks that an average 10 year old could. I don't consider that generally intelligent.

but it can perform all kinds of tasks. generally intelligent isn't universally intelligent, is it?

useful, sure. like a calculator for text. intelligent? no, and neither is a calculator

Yes, and "that question is in not-even-wrong territory", respectively.

a question, especially this one, can't really be "not even wrong"

Or just a realization that they are mainly up against google with a massive money factory. They need their own money factories or they won’t survive long term

AGI is the ultimate money factory if they could deliver it on the timelines that they were projecting until recently.

Its impossible to know, they are in a longer fight with googs

I disagree for two reasons.

1) the two decisions do not seem related to each other. OpenAI has capital to spend and is seeking distribution methods to shore up continued access to future capital. That strategic decision seems totally unrelated to their estimated timelines for when AGI (whatever definition you are using) will show up. Especially because they are in a race against other players. It may be a soft signal that more capital is not going to speed up the AGI timeline right now, but even that is a soft signal.

2) I think we already have AGI for any reasonable definitions of the terms 'artificial' 'general' and 'intelligence'. To wit: I can ask Gemini 2.5 a question about basically anything, and it will respond more coherently and more accurately than the vast majority of the human population, about a vast array of subjects.

I do not understand what else AGI could mean.

(In case it matters, I am also an AI researcher, I know many AI researchers, and many-but-not-all agree with me)

I asked Gemini to read a clock for me with hands on 10 and 2 and it got the long hand and the short hand backwards, probably because of the massive trove of online documentation about the symmetry of 10 after 10 being aesthetically pleasing for PR materials and icons or some such nonsense unrelated to the question or the clock.

I don't know about you, but I learned how to read an analog clock in kindergarten and Gemini got it wrong.

Sorry, so what? A few different ways to respond to this:

1) Please do me a favor and take the GPQA benchmark. I'm curious to see how you would do. Now go find the nearest kindergartner and ask them to take it. Curious to see how they would do. Maybe random 'ha gotcha!' tasks are not good measures of intelligence?

2) Depending on how you want to measure, the average human is ingesting somewhere between 10 and 100 mb per second. By the time you were in kindergarten (5yo) you would have ingested, conservatively, nearly 2 petabytes of highly multimodal data. Meanwhile, you are comparing against a system that has to understand everything it knows about the world from text (to a first approximation).

3) It seems very strange that reading a clock is a measure of intelligence at all. Unless you think large parts of GenZ are simply not generally intelligent

re "agree with"; You sound like a moron. How much $ are AI researchers who have dedicated their life to it gonna make by shorting AI? If you can't even solve this much then you don't have well-rounded intelligence. let alone be an AI researcher.

Also, do you even know what General mean? Gemini can't even tell me what time the library is open today, while even a 3y kid can. So much for "accurately".

If agi is coming, or even another ai as overwhelming as chatgpt is to its prior age, Then investing in all those companies is the Last thing to do. since they'd be leapfrogged by what's coming.

By investing in them one declares that there are no leapfrogs coming. Aka no agi, or even anything close to 10x chatgpt.

With that therefore, the battlefield shifts to being the best middleman. hence all those senseless amounts of money thrown around. For the masses will no longer need to personally seek out God Altman for their top oracling needs, and so someone can come between God and man, capture all value like microsoft did to ibm, and use it to compete building a new God (read: new scam). Rinse repeat.

There's no post-LLM story. This consolidation is both building product and user-capture before that becomes wildly admitted.

I think LLMs will become more useful and more efficient over time as models refine but these aren't the (AI) droids you're looking for.

what does LLVM mean here?

I assume it's 'large language and vision model'.

But, for sanity's sake, if we insist on putting 'vision' in there, let's at least call them LVLMs!

I wasn't trying to coin a new term. Genuine mistake!

like when Linda McMahon called AI A1

I can't wait for AG1 so we can all quit our jobs as we are replaced by AG1

Probably LLM.

If we’re talking about the singularity robot takeover fast takeoff, maybe that’s true.

But Sam and others have said they see AGI is an uneven process that may not have a clear finish line. The intelligence is spiky and some parts will be superhuman while other parts lag.

Note that they started saying that recently after their earlier projections didn't pan out. The "uneven process without a clear finish line" angle was Altman recently trying to reset expectations, which means it doesn't contradict OP's thesis that this move towards product is further admission that AGI is going to be much messier than they initially predicted.

That sounds exactly what you would say if you had staked hundreds of millions of dollars and your personal reputation on something you increasingly know isn't possible.

Sounds like excuses

Can't do a singularity robot takeover without robots and you may as well use designer ones.

Dude just says whatever he thinks will make Line Go Up on his accounts. It's basically the only thing he does.

"Make Line Go Up, but without humans" might be the best definition of AGI i've seen so far

Doesn't this contradict the whole idea of AGI? How is it still "general" if it is "spiky" with numerous gaps in its "intelligence"?

It sounds like a CEO moving the goalposts when asked to accomplish something they don't think they can deliver.

Just means by the time you get agreement that every checkbox has been checked, much of the world will already have been dominated by the spiky parks to AI long before.

We'll be living in a mostly AGI-ish world long before it gets declared. People might not even care about declaring it at that point.

Because LLM's as they exist right now are incredibly useful and you can make a lot of money from them? AGI isn't god. It might not even be especially useful.

Is there a profitable LLM out there right now that anyone can point to as an example?

it's funny I feel like top of the line LLMs are basically AGI already or very close, you can have reasonable discussions with them about any subject etc & that a lot of anti LLM talk is grasping at straws & goalpost moving

People's definitions of AGI vary a lot, although I agree LLMs are getting pretty good.

RE: product-focused moves

These are tentacles that AGI will need.

Yeah, AGI needs a VS Code fork.

Reply on news.ycombinator.com