Keep pushing it down the road until you realize it's a pipe dream problem.
AGI is an engineering problem, not a model training problem
141–150 of 442 posts
Re: AGI is an engineering problem, not a model training problem
#142Earlier quoted context omitted.
It is especially not obvious because this was written using ChatGPT-5. One appreciates the (deliberate?) irony, at least. (Or at least, surely if they had asymptoted, OP should've been able to write this upvoted HN article with an old GPT-4, say...)
> this was written using How do you know?
My favorite tell-tale sign:
> The gap isn’t just quantitative—it’s qualitative.
> LLMs don’t have memory—they engage in elaborate methods to fake it...
> This isn’t just database persistence—it’s building memory systems that evolve the way human memory does...
> The future isn’t one model to rule them all—it’s hundreds or thousands of specialized models working together in orchestrated workflows...
> The future of AGI is architectural, not algorithmic.
Re: AGI is an engineering problem, not a model training problem
#143It's a research problem, a science problem. And then an engineering problem to industrialize it. How can we replicate intelligence if we don't even know how it emerges from our brains?
We do not «replicate». We implemented computing without any need of a brain-neural theory of arithmetic.
At its core, arithmetic is a deterministic set of rules that can be implemented with logic gates. Computing is just taking that and scaling it up a billion times. What is intelligence? How do you implement intelligence if nobody can provide a consistent, clear definition of what it is?
Re: AGI is an engineering problem, not a model training problem
#144Earlier quoted context omitted.
We don’t need such a definition of general intelligence to conclude that biological humans have it, so I’m not sure why we’d such a definition for AGI.
I disagree. We claim that biological humans have general intelligence because we are biased and arrogant, and experience hubris. I'm not saying we aren't generally intelligent, but a big part of believing we are is because not believing so would be psychologically and culturally disastrous. I fully expect that, as our attempts at AGI become more and more sophisticated, there will be a long period where there are inte…
I can say it. Humans are not "generally intelligent". We are intelligent in a distribution of environments which are similar enough to ones we are used to. There's no way to be intelligent with no priors on environment basically by information theory (you can make your environment to be adversarial to the learning efficiency in "intelligent" beings which comes from priors)
Re: AGI is an engineering problem, not a model training problem
#145Earlier quoted context omitted.
We don’t need such a definition of general intelligence to conclude that biological humans have it, so I’m not sure why we’d such a definition for AGI.
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…
How do we know when a newborn has achieved general intelligence? We don't need a definition amenable to proof.
Re: AGI is an engineering problem, not a model training problem
#146The forgone conclusion that LLMs are the key or even a major step towards AGI is frustrating. They are not, and we are fooling ourselves. They are incredible knowledge stores and statistical machines, but general intelligence is far more than these attributes.
My thoughts are that LLMs are like cooking a chicken by slapping it: yes, it works, but you need to reach a certain amount of kinetic energy (the same way LLMs only "start working" after reaching a certain size).
So then, if we can cook a chicken like this, we can also heat a whole house like this during winters, right? We just need a chicken-slapper that's even bigger and even faster, and slap the whole house to heat it up.
There's probably better analogies (because I know people will nitpick that we knew about fire way before kinetic energy), so maybe AI="flight by inventing machines with flapping wings" and AGI="space travel with machines that flap wings even faster". But the house-sized chicken-slapper illustrates how I view the current trend of trying to reach AGI by scaling up LLMs.
Re: AGI is an engineering problem, not a model training problem
#147Earlier quoted context omitted.
On the contrary, we have one working example of general intelligence (humans) and zero of quantum computing.
Do we have a specific enough definition of general intelligence that we can exclude all non-human animals?
No one said "exclusively humans", and that's not relevant.
Re: AGI is an engineering problem, not a model training problem
#148Earlier 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?”
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.
Re: AGI is an engineering problem, not a model training problem
#149Earlier quoted context omitted.
Why does it need to exclude all non human animals? Could it not be a difference of degree rather than of kind?
The post I was responding to had > On the contrary, we have one working example of general intelligence (humans) I think some animals probably have what most people would informally call general intelligence, but maybe there’s some technical definition that makes me wrong.
If some animals also have general intelligence then we have more than one example, so this simply isn't relevant.
Re: AGI is an engineering problem, not a model training problem
#150If you believe the bitter lesson, all the handwavy "engineering" is better done with more data. Someone likely would have written the same thing as this 8 years ago about what it would take to get current LLM performance. So I don't buy the engineering angle, I also don't think LLMs will scale up to AGI as imagined by Asimov or any of the usual sci-fi tropes. There is something more fundamental missing, as in missing…
What will it scale up to if not AGI? OpenAI has a synthetic data flywheel. What are the asymptotics of this flywheel assuming no qualitative additional breakthrough?
It's established science (see Chomsky) that a probablistic model will never achieve something approaching AGI