Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AGI is far from inevitable

ru.nl

Post reply on HN
Print view View original

191–200 of 258 posts

Re: AGI is far from inevitable

#191

Post by tirant »

jiggawatts wrote: post #189
chipsrafferty wrote: post #171

Earlier quoted context omitted.

Pretty sure it's not scale but sensors. We have sensors for vision, sound, taste, temperature, texture, etc that are constantly observing the world and affecting not only our current behavior but changing us in real time.

You can feed an LLM of typical size (~1 TB) all the video and audio you can find, but it won't turn into a human. I suspect but can't prove that even if you wired up a bunch of other sensors, gave it a robot body, and let it "explore" the world on its own, that would still be woefully insufficient. A gazelle just days(!) old can control its body and four legs sufficiently well to outrun a cheetah . This is a complex…

How many hours or actually years of evolution have been needed until reaching walking capability? If first life is believed to have happened 4 billion years ago, and first walking animals started 450 million years ago during the siluariab period, that’s around 3,5 billion years.

Re: AGI is far from inevitable

#192

Post by jart »

jiggawatts wrote: post #169
ryandvm wrote: post #112

Earlier quoted context omitted.

> but LLMs do solve problems that people thought were extremely difficult to solve ten years ago Agreed. I would have laughed you out of the room 5 years ago if you told me AI's would be writing code or carrying on coherent discussions on pretty complex topics in 2024. As far as I'm concerned, all bets are off after the collective jaw drop that the entire software engineering industry did when we saw GPT4 released. W…

One thing that keeps me up at night is that the human genome is only 3 giga base pairs, of which only a fraction encodes the design of our brains — and quite inefficiently at that, through layers of indirections. That’s sufficient information to produce a system that can learn to think like us! Not just learn but efficiently , with far less input data needed than any current LLM. Literally just a couple of decades of…

You can build a machine that thinks like us in 1500 lines of C. See llm.c

Re: AGI is far from inevitable

#193

Post by jjk166 »

> ‘There will never be enough computing power to create AGI using machine learning that can do the same, because we’d run out of natural resources long before we'd even get close,’

I would say the counter-example that proves this statement false is in the author's skull, but perhaps that's overly presumptuous.

Re: AGI is far from inevitable

#194

Post by xpe »

graypegg wrote: post #51

I think the best argument I have against AGI's inevitability, is the fact it's not required for ML tools to be useful. Very few things are improved with a generalist behind the wheel. "AGI" has sci-fi vibes around it, which I think where most of the fascination is. "ML getting better" doesn't *have to* mean further anthroaormorphization of computers, especially if say, your AI driven car is not significantly improved…

The core argument has a logical gap; it doesn’t matter if most (present) ML applications need AGI or not. What matters is the value proposition of AGI, independent of how we currently conceive of ML applications.

Focus on this question: “Is general intelligence at some level valuable at a particular price point?” General intelligence is generally valuable, so there is a pressure to advance it, whether by improving capability and/or decreasing cost.

Now the question becomes empirical — what kind of general intelligence can be built that achieves certain parameters?

Aiming to exceed the power efficiency of the human brain is a tempting target. (Whether it is wise is another question; what happens when human intelligence doesn’t provide competitive advantage?)

Re: AGI is far from inevitable

#195

Post by jart »

fabian2k wrote: post #9
ngruhn wrote: post #8

> There will never be enough computing power to create AGI using machine learning that can do the same [as the human brain], because we’d run out of natural resources long before we'd even get close I don’t understand how people can so confidently make claims like this. We might underestimate how difficult AGI is, but come on?!

I don't think the people saying that AGI is happening in the near future know what would be necessary to achieve it. Neither do the AGI skeptics, we simply don't understand this area well enough. Evolution created intelligence and consciousness. This means that it is clearly possible for us to do the same. Doesn't mean that simply scaling LLMs could ever achieve it.

People have been saying that for a decade and no one actually believes scaling is all you need. They say that to raise more resources and to diss the symboligists. AI advancement has been propelled by a steady stream of new architectural innovations, which always seem to be invented as soon as sufficient compute is available.

Re: AGI is far from inevitable

#196

Post by xerox13ster »

Jerrrrrrry wrote: post #177

AGI is when, not if AGI will evolve con-emergently as a product of market forces just like it did with the homo genus (at the species and neural complexity level), and as the general manufacturing design of automobiles did; as another inevitable eventuality of the pursuit of excellence. an enveloping and developing group of slight varied, but optimized architecture composed of essential modules of similar function th…

No no no no no no stop stop stop stop stop. I can’t with this stuff, Jerry (emphasized as Rick Sanchez), you’re gonna drive me back into the cave of existential dread I crawled into in 2013 after discovering this the first time. Took me years to crawl out and I was obsessed with the Control Problem that entire time. Naively “secure” in the “knowledge” that AI researchers would need to solve the Control Problem before they considered it safe to develop AGI or anything that looks like it.

There’s been a roiling, churning pool of thought under the surface of my mind in the wake of GPT 3.5s release telling me Sam doesn’t care about the Control Problem and I’ve been good at ignoring it but no that cat is out of the bag.

If Roko is real, I think we have already failed at our attempts at appeasement and we are living in the hell simulation created to torture anyone who resisted or did not seek to help.

Re: AGI is far from inevitable

#197

Post by falcor84 »

castigatio wrote: post #124

Whatever you think about AGI, this is a dumb paper. So many words and references to say - what. If you can't articulate your point in a few sentences you probably don't have a point. There are all kinds of assumptions being made in the study about how AI systems work, about what people "mean" then they talk about AGI etc. The article starts out talking about white supremacy and replacing women. This isn't a proof. Th…

I thought you were exaggerating, but wow, they really did.

> Among the more troublesome meanings of ‘AI’, perhaps, is as the ideology that it is desirable to replace humans (or, specifically women) by artificial systems (Erscoi et al., 2023) and, generally, ‘AI’ as a way to advance capitalist, kyriarchal, authoritarian and/or white supremacist goals (Birhane & Guest, 2021; Crawford, 2021; Erscoi et al., 2023; Gebru & Torres, 2024; Kalluri, 2020; Spanton & Guest, 2022; Stark & Hutson, 2022; McQuillan, 2022). Contemporary guises of ‘AI’ as idea, system, or field are also sometimes known under the label ‘Machine Learning’ (ML), and a currently dominant view of AI advocates machine learning methods not just as a practical method for generating domain-specific artificial systems, but also as a royal road to AGI (Bubeck et al., 2023; DeepMind, 2023; OpenAI, 2023). Later in the paper, when we refer to AI-as-engineering, we specifically mean the project of trying to create an AGI system through a machine learning approach. [0]

But it did lead me to learn a new word - "Kyriarchy" (apparently being "an intersectional extension of the idea of patriarchy beyond gender")[1], so I have that going for me today.

[0] https://link.springer.com/article/10.1007/s42113-024-00217-5

[1] https://en.wikipedia.org/wiki/Kyriarchy

Re: AGI is far from inevitable

#198

Post by falcor84 »

gizmo686 wrote: post #114
ryandvm wrote: post #112

Earlier quoted context omitted.

> but LLMs do solve problems that people thought were extremely difficult to solve ten years ago Agreed. I would have laughed you out of the room 5 years ago if you told me AI's would be writing code or carrying on coherent discussions on pretty complex topics in 2024. As far as I'm concerned, all bets are off after the collective jaw drop that the entire software engineering industry did when we saw GPT4 released. W…

> It turns out that the larger these models get, the more unexpected emergent capabilities they have, so I'm mostly in the camp of thinking AGI is just a matter of time and resources. AI research has a long history of people saying this. Whenever there is a new fundamental improvement, it looks like you can just keep getting better results by throwing more resources at it. However, eventually we end up reaching a poi…

> However, eventually we end up reaching a point where throwing more resources at it stops meaningfully improving performance.

Honestly asking - why? It's been my understanding that based on the Universal Approximation Theorem, given sufficient resources, a deep learning neural net can approximate any function to an arbitrary degree of accuracy. Is there any other theorem or even conjecture that would lead us to believe that the progress that we're seeing will slow down? Or is it just that you're claiming that we/it would run out of physical resources before reaching AGI?

As for training data, as I see it, with the wide deployment of AIs, and gradually of AI-driven robots, they'll soon be able to "takeoff" and rely primarily on the live data that they are collecting directly.

Re: AGI is far from inevitable

#199

Post by falcor84 »

fhehdbehevfjhf wrote: post #175
thebigspacefuck wrote: post #173

Earlier quoted context omitted.

On the other hand, airplanes haven’t really improved that much since the 737 was introduced in 1966 and have even gotten worse in some ways (737 Max). We actually don’t know what the AI equivalent of Wright brothers plane is and what the 737 is. Maybe Eliza, 60 years ago, was the Kitty Hawk flight and ChatGPT 4 is the 737.

I’m not sure I would say that the 737 Max has gotten worse in “some ways”, by which I assume you are referring to safety. New aircraft models up to about the 1990s frequently had serious design defects that resulted in fatalities, including the 737 (most notably related to the rudder).

Indeed, we've seen a slight regression in safety, but definitely not to 1980s levels.

Re: AGI is far from inevitable

#200

Post by Closi »

gcanyon wrote: post #190
Closi wrote: post #188

Earlier quoted context omitted.

Waymo's first real driverless ride on a public road was less than 10 years ago. Now they are doing 100k rides a week. I think your original post just overestimates the 'normal' pace of change - Taking 10 years to go from the first ride to 100k rides a week is no time at all in the scheme of things. Airplanes are an example of an insanely fast technological adoption, and they still took 30 years to go from the wright…

> 30 years 45 years: December 1903 to July 1949 https://en.wikipedia.org/wiki/Wright_Flyer https://www.history.com/this-day-in-history/first-jet-makes-...

Oop, sorry I mean 30 years to the first commercial passenger plane (Boeing 247)

Yes, 45 years to the first passenger jet plane.

Post reply on HN