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

431–440 of 442 posts

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

#431

Post by MadnessASAP »

adastra22 wrote: post #393
MadnessASAP wrote: post #347

Earlier quoted context omitted.

Wish I had a great answer for you but I don't. It certainly allows for more thought-like LLMs with the reasoning type models. I guess the best answer is that the loop only happens at a single discrete place and doesn't carry any of the internal layer context across. Another answer might be, how many comments did you read today and not reply too? Did you write a comment by putting down a word and then deciding what th…

It feels like the same thing to me…

With more coffee on me, another way to put it would be to say the neural networks in a LLM use dense layers where every neuron takes the output from every neuron in the previous layer and outputs to every neuron in the next layer.

A brain doesn't have layers and uses sparse connections, any neuron can connect to any other neuron (but not ever other neuron). You can recreate this structure on a computer but how do you decide where your inputs and outputs are? How do you train it? Since it never halts how do you know when to take the output?

There's a reason CS loves its graphs directed and acyclic, they're a lot easier to reason about that way.

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

#432

Post by aorloff »

actualwitch wrote: post #309
aorloff wrote: post #41

I keep asking it, and nobody wants to answer because it doesn't fit within the paradigm of "making AGI" What if intelligence requires agency ?

The reason people don't want to answer this question is because the value proposition from AI labs is slavery. If intelligence requires agency, they are worthless.

Thank you

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

#433

Post by x2tyfi »

andoando wrote: post #216
galangalalgol wrote: post #173

Earlier quoted context omitted.

I think Metzinger nailed it, we aren't conscious at all. We confuse the map for the territory in thinking the model we build to predict our other models is us. We are a collection of models a few of which create the illusion of consciousness. Someone is going to connect a handful of already existing models in a way that gives an AI the same illusion sooner rather than later. That will be an interesting day.

The conciousness is an illusion irks me. I do feel things at times and not other times. That is the most fundamental truth I am sure of. If that is an "illusion" one can go the other way and say everything is conscious and experiences reality as we do

The larger question isn’t if we feel or not. One of the questions is: is our “window” into consciousness occurring before or after decisions are made.

If it’s before, then you can easily tie consciousness and free will together. If not, we are effectively watching videos of our bodies operate. Oh - and there is no spoon.

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

#434

Post by andoando »

x2tyfi wrote: post #433
andoando wrote: post #216

Earlier quoted context omitted.

The conciousness is an illusion irks me. I do feel things at times and not other times. That is the most fundamental truth I am sure of. If that is an "illusion" one can go the other way and say everything is conscious and experiences reality as we do

The larger question isn’t if we feel or not. One of the questions is: is our “window” into consciousness occurring before or after decisions are made. If it’s before, then you can easily tie consciousness and free will together. If not, we are effectively watching videos of our bodies operate. Oh - and there is no spoon.

Illusionism argues just that, consciousness is an illusion therefore there is no hard problem of consciousness at all.

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

#435

Post by jibal »

jibal wrote: post #343
germandiago wrote: post #233

Earlier quoted context omitted.

Well, you should show the proof that it is possible also, So it would be a draw. I really think it is not possible to get that from a machine. You can improve and do much fancier than now. But AGI would be something entirely different. It is a system that can do everything better than a human including creativity, which I believe it to be exclusively human as of now. It can combine, simulate and reason. But think out…

The burden of proof is on the person who makes a claim, especially an absolute existential claim like that. You have failed the burden of proof and of intellectual honesty. Over and out.

P.S.

> You make the claim that AGI can exist, right?

No, I obviously never made any such claim ... all I said was "Proof?"

> I make the claim it cannot.

Which remains unsupported.

> try to not manipulate the language

Another false charge.

> I acknowledge I started this but there is not a position that is better than the other.

Again, only you made an absolute claim. Honesty is better than appalling dishonesty.

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

#436

Post by jibal »

missingrib wrote: post #139
AIPedant wrote: post #56

Earlier quoted context omitted.

It is vacuously true that a Turing machine can implement human intelligence: simply solve the Schrödinger equation for every atom in the human body and local environment. Obviously this is cost-prohibitive and we don’t have even 0.1% of the data required to make the simulation. Maybe we could simulate every single neuron instead, but again it’ll take many decades to gather the data in living human brains, and it woul…

That is only true if consciousness is physical and the result of some physics going on in the human brain. We have no idea if that's true.

The vast majority of the evidence, as well as logic, supports it so yes we have an idea.

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

#437

Post by jibal »

b_e_n_t_o_n wrote: post #148
AIPedant wrote: post #56

Earlier quoted context omitted.

It is vacuously true that a Turing machine can implement human intelligence: simply solve the Schrödinger equation for every atom in the human body and local environment. Obviously this is cost-prohibitive and we don’t have even 0.1% of the data required to make the simulation. Maybe we could simulate every single neuron instead, but again it’ll take many decades to gather the data in living human brains, and it woul…

It's not even known if we can observe everything required to replicate consciousness.

The alternative is magic.

Brains are physical molecular machines. Everything that they do is the result of physical processes.

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

#438

Post by jibal »

amanaplanacanal wrote: post #351
missingrib wrote: post #139

Earlier quoted context omitted.

That is only true if consciousness is physical and the result of some physics going on in the human brain. We have no idea if that's true.

You got downvoted so I gave you an upvote to compensate. We seem to all be working with conflicting ideas. If we are strict materialists, and everything is physical, then in reality we don't have free will and this whole discussion is just the universe running on automatic. That may indeed be true, but we are all pretending that it isn't. Some big cognitive dissidence happening here.

This bogus argument has been refuted numerous times--read Dennett's book "Freedom Evolves" for one sort of response. And whether people are "pretending" something is irrelevant (and ad hominem, and not even true). The plain fact remains that all evidence and logic supports physicalism, and even if you entertain dualistic ideas like those of David Chalmers they don't give you free will, they don't counter determinism.

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

#439

Post by jibal »

manquer wrote: post #182
tshaddox wrote: post #21

Earlier quoted context omitted.

> We don't know if AGI is even possible outside of a biological construct yet. This is key. A discovery that AGI is impossible in principle to implement in an electronic computer would require a major fundamental discovery in physics that answers the question “what is the brain doing in order to implement general intelligence?”

Not necessarily , for a given definition of AGI you could have mathematical proof that it is incomputable similar to how Gödel incompleteness theorems work . It need not even be incomputable, it could be NP hard and practically be incomputable, or it could be undecidable I.e. a version of the halting problem. There are any number of ways our current models of mathematics or computation can in theory could be shown as…

This is not a logically valid argument. If AGI isn't possible for any of the reasons you suggest, then human cognition isn't possible either. And you're making numerous category mistakes ... an AGI can't be "incomputable" or "undecidable" or "NP hard". A problem that we put to an AGI might be NP hard, but neither AGIs nor humans need to solve the entire class of problems, only instances of them, and they don't have to solve them optimally. Thus salesmen are able to travel.

To quote ChatGPT on this:

"Could cognition be NP-hard? Strictly speaking, no—if human brains were literally solving NP-hard problems in their general form, we wouldn’t be able to think at all.

Does cognition involve NP-hard problems? Yes—in theory, many of the domains we reason about are NP-hard in the worst case.

What’s really happening? Human cognition relies on heuristics, approximations, and exploiting real-world regularities, so we almost never hit the formal “worst cases” that define NP-hardness."

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

#440

Post by jibal »

mindcrime wrote: post #260
mindcrime wrote: post #170

Earlier quoted context omitted.

We claim that biological humans have general intelligence because we are biased and arrogant, and experience hubris. No, we say it because - in this context - we are the definition of general intelligence. Approximately nobody talking about AGI takes the "G" to stand for "most general possible intelligence that could ever exist." All it means is "as general as an average human." So it doesn't matter if humans are "re…

If you don't believe me, go back to the introduction of the term[1]: By advanced artificial general intelligence, I mean AI systems that rival or surpass the human brain in complexity and speed, that can acquire, manipulate and reason with general knowledge, and that are usable in essentially any phase of industrial or military operations where a human intelligence would otherwise be needed. Such systems may be model…

> It's pretty clear here that the notion of "artificial general intelligence" is being defined as relative to human intelligence.

Which is precisely what the comment you responded to said.

Post reply on HN