Markov Chains are the Original Language Models

If you mean something projecting itself into the air, then yes. Rockets fly, Kites fly, bullets fly, planes too fly.

But if by flight you mean replicating what birds do to get into the sky & come down. Rockets, Kites, bullets & planes don't do that.

Yes, penguins don't swim like sardines either and snakes don't "walk" like apes.

> Why is the bird privileged here ?

There is no privilege, I was just using it as an example to illustrate that what human beings do is different from what LLMs do, even though they are all called "intelligence".

Well scientists do. If your to listen to talks from Yan Lecun & Geoff Hinton or Chomsky. They do care and the do want to figure it out.

> Nobody is trying to build a digital human here so, “they’re not doing what humans do” is about as meaningless as it gets.

Actually alot of AI researchers are trying todo this. It's just that we haven't yet made any significant strides in the field. It is no coincidence that neural networks are loosely inspired by nature.

The original goal of AI research was to create a "thinkin machine" with intelligence equivalent to that of a human being.

The value of scientific insights & industrial applications this would make possible are obvious.

There haven't been any significant breakthroughs in AI comparable to human intelligence.

Industry has simply taken up auxiliary research for applications. But the core scientific goal has not yet been achieved.

>I mean sure, you could cook up whatever definition of intelligence you like, make it so that humans are the only things that pass just because. Then, what's the point? It's not a useful definition

Intelligence is a very wide spectrum. Computers, calculators, do have something that can be classified as a form of intelligence. But it is more important to understand what they are not, i.e intelligence fully that satisfies what humans are capable of.

I mean if someone says they "do drugs".

What do they mean? The drink coffee, tea, smoke tobacco, codine, weed or heroin. A pharmacist a law enforcement person, and a highschool kid will have very different classifications on what a drug really is.

That is why we need to "cook up definitions" so that we know exactly what someone is talking about. There is no word for "human specific intelligence"

And the wright brothers sought to create a flying machine equivalent to that of a bird. They studied birds too.

That doesn't mean they wouldn't have taken, "but but it has no feathers, it doesn't flap !!. It's only pretend flying" with anything more than a laugh to your face.

People are trying to build intelligence equivalent to a human being, not a human being.

You are not Turing complete. We do like to pat ourselves on the back and say, "Humans? Of course, they're Turing complete" but you're not. You do not have infinite memory either theoretically (all else fails, you'll die) or practically (boredom, lack of concentrating, lack of interest, and a memory that is very prone to making stuff up or discarding vital details at a disturbing frequency).

In fact, you are a very poor excuse for a Turing machine. Simulate any computation? You wish you could. Your brain is a finite state machine through and through.

So why do we continue to deceive and pat ourselves on the back ?

Because, well besides the dash of human exceptionalism and narcissism we're so well known for, limited =/ trivial.

You see, I don't really care how precisely Transformers meet some imaginary goal humans don't meet themselves.

It's not important how Turing complete transformers are, only that they could potentially learn any class of computations necessary via training.

>Moreover adding memory is no longer the simple "forward pass" system that you were arguing had no significant computational limitations

Memory or not, all the computation is still being performed just in the forward pass.

>Second, and more importantly, the fact that through clever human manipulation you can express a Turing machine using an LLM does not mean that such a machine is learned through gradient descent training.

It also doesn't mean it couldn't be.

>There is no basis to the claim that today's systems have converged on model weights that implement "higher-order" computation.

Results are basis enough. If I say, "you're intelligent", it's because you appear to be so. It's an assumption, not truth I've verified after peeking into your brain. All the properties I might ascribe to humans: 'intelligence', 'conciousness' are all assumptions I ascribe based on the results I see and nothing else. I have no proof you are performing any 'higher order computation' either.