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

AGI is an engineering problem, not a model training problem

vincirufus.com

Post reply on HN
Print view View original

271–280 of 442 posts

Re: AGI is an engineering problem, not a model training problem

#271

Post by Jensson »

fastball wrote: post #113
Fricken wrote: post #95

Earlier quoted context omitted.

We're fixated on human intelligence but a computer cannot even emulate the intelligence of a honeybee or an ant.

How do you mean? AFAICT computers can definitely do that. Sure, it won't be the size of an ant, but we definitely have models running on computers that have much more complexity than the life of an ant.

> Sure, it won't be the size of an ant, but we definitely have models running on computers that have much more complexity than the life of an ant.

Do we? Where is the model that can run an ant and navigate a 3d environment, parse visuals and different senses to orient itself, figure out where it can climb to get to where it needs to go. Then put that in an average forest and navigate trees and other insects and try to cooperate with other ants and find its way back. Or build an anthill, an ant can build an anthill, full of tunnels everywhere that doesn't collapse without using a plan.

Do we have such a model? I don't think we have anything that can do that yet. Waymo is trying to solve a much simpler problem and they still struggle, so I am pretty sure we still can't run anything even remotely as complex as an ant. Maybe a simple worm, but not an ant.

Re: AGI is an engineering problem, not a model training problem

#272

Post by Jensson »

dragonwriter wrote: post #257
Fricken wrote: post #256

Earlier quoted context omitted.

Having aptitude in mathematics was once considered the highest form of human intelligence, yet a simple pocket calculator can beat the pants off most humans at arithmetic tasks. Conversely, something we regard as simple, such as selecting a key from a keychain and using to unlock a door not previously encountered is beyond the current abilities of any machine. I suspect you might be underestimating the real complexit…

> Having aptitude in mathematics was once considered the highest form of human intelligence, yet a simple pocket calculator can beat the pants off most humans at arithmetic tasks. Mathematics has been a lot more than arithmetic for... a very long time.

But arithmetics was seen as requiring intelligence, as did chess.

Re: AGI is an engineering problem, not a model training problem

#273

Post by Jensson »

Sevii wrote: post #192

I don't understand how people feel comfortable writing 'LLMs are done improving, this plateau is it.' when we haven't even gone an entire calendar year without seeing improvements to LLM based AI.

You can expand to cover more of a plateau, that is still an improvement but also still a plateau as you aren't going higher.

So models improve on specific tasks, but they don't really improve generally across the board any longer.

Re: AGI is an engineering problem, not a model training problem

#274

Post by Jensson »

JSR_FDED wrote: post #213
dsign wrote: post #177

I have colleagues that want to plan each task of the software team for the next 12 months. They assume that such a thing is possible, or they want to do it anyway because management tells them to. The first would be an example of human fallibility, and the second would be an example of choosing the path of (perceived) least immediate self-harm after accounting for internal politics. I doubt very much we will ever bui…

Interesting, I hadn’t thought about it that way. But can a thing on the other end of an API call ever truly have a “stake“?

> But can a thing on the other end of an API call ever truly have a “stake“?

That is their goal function they are trained for, it is like dopamine and sex for humans they will do anything to get it.

Re: AGI is an engineering problem, not a model training problem

#275

Post by exe34 »

frabcus wrote: post #202
EMIRELADERO wrote: post #184

Earlier quoted context omitted.

I don't see how your explanation leads to consciousness not being a thing. Consciousness is whatever process/mechanisms there are that as a whole produce our subjective experience and all its sensations, including but not limited to touch, vision, smell, taste, pain, etc.

You've missed our consciousness of our inner experiences. They are more varied than just perception at the footlights of our consciousness (cf Hurlburt): Imagination, inner voice, emotion, unsymbolized conceptual thinking as well as (our reconstructed view of our) perception.

oh no, those people without an inner voice are now cowering in a corner...

Re: AGI is an engineering problem, not a model training problem

#277

Post by bubblyworld »

crystal_revenge wrote: post #190
ants_everywhere wrote: post #90

Earlier quoted context omitted.

> there is missing philosophy I doubt it. Human intelligence evolved from organisms much less intelligent than LLMs and no philosophy was needed. Just trial and error and competition.

The magical thinking around LLMs is getting bizarre now. LLMs are not “intelligent” in any meaningful biological sense. Watch a spider modify its web to adapt to changing conditions and you’ll realize just how far we have to go. LLMs sometimes echo our own reasoning back at us in a way that sounds intelligent and is often useful, but don’t mistake this for “intelligence”

Watch a coding agent adapt my software to changing requirements and you'll realise just how far spiders have to go.

Just kidding. Personally I don't think intelligence is a meaningful concept without context (or an environment in biology). Not much point comparing behaviours born in completely different contexts.

Re: AGI is an engineering problem, not a model training problem

#278

Post by dns_snek »

idiotsecant wrote: post #243
bsenftner wrote: post #134

Earlier quoted context omitted.

Are you sure? Do you require dozens, to hundreds, to thousands of examples before you understand a concept? I expect no. That is because you have comprehension that can generalize a situation to basic concepts which you apply to other situations without effort. You comprehend. AI cannot do that: get the idea from a few, under a half dozen examples if necessary. Often a human needs 1-3 examples before they can general…

You spend every waking minute for 20 years or so accumulating training data. You don't learn addition and then independently discover vector calculus.

Individual people don't but we did it as a species. Any purported AGI should be capable of doing the same.

Re: AGI is an engineering problem, not a model training problem

#279

Post by Jensson »

tshaddox wrote: post #93
bastawhiz wrote: post #85

Earlier quoted context omitted.

Well general intelligence in humans already exists, whereas general intelligence doesn't yet exist in machines. How do we know when we have it? You can't even simply compare it to humans and ask "is it able to do the same things?" because your answer depends on what you define those things to be. Surely you wouldn't say that someone who can't remember names or navigate without GPS lacks general intelligence, so it's…

> You can't even simply compare it to humans and ask "is it able to do the same things?" because your answer depends on what you define those things to be. Right, but you can’t compare two different humans either. You don’t test each new human to see if they have it. Somehow we conclude that humans have it without doing either of those things.

> You don’t test each new human to see if they have it

We do, its called school and we label some humans with different learning disabilities. Some of those learning disabilities are grave enough that they can't learn to do tasks we expect humans to be able to learn, such humans can be argued to not posses the general intelligence we expect from humans.

Interacting with an LLM today is like interacting with an Alzheimer patient, they can do things they already learned well but poke at it and it all falls apart and they start repeating themselves, they can't learn.

Re: AGI is an engineering problem, not a model training problem

#280

Post by artninja1988 »

nativeit wrote: post #92

Say we get there--all the way, full AGI, indistinguishable from conscious intelligence. Congratulations, every thing you do from that point on (very likely everything from well before that point) that is not specifically granting it free will and self-determination is slavery. Doesn't really feel like a real "win" for any plausible end goal. I'm really not clear on why anyone thinks this is a good idea, or desirable,…

I don't think agi necessarily implies consciousness, at least many definitions of it. OpenAIs definition is just ai that does most of economically viable work
Post reply on HN