Merged Artificial Intelligence

[Dr.Sid]

Computers were not better at theoretical math than humans. Not before last week.
I mean it's still questionable, but the fact you can just throw compute at the problem, and get a solution, is major shift. And considering how few people is doing theoretical math, and often not even full time, OpenAI advanced math by decades of current speed. Mathematicians are panicking a bit .. but us programmers are experiencing this for years now. They will get used to it. They will realize so far somebody has to formulate those problem, be interested in the solution, and make youtube videos about them. It's simply their turn now. Who will be next ?

 

[W.D.Clinger]

He says some interesting things. I'm unsure though. He thinks this will discourage a 17-year-old who is good at math from pursuing a career in advanced math. Eventually we may reach a point where nobody goes into the field because computers are so much better at it than humans, and we could reach the point where we've basically outsourced our higher thinking to machines.

Whether you agree or disagree, I think it's at least interesting to hear his argument.

It is certainly an interesting argument, worthy of concern.

Computers were not better at theoretical math than humans. Not before last week.

That's like saying computers were not better at chess than humans. Not before 1997.

In truth, however, the best computer chess programs were better at chess than most humans long before DEEP BLUE defeated then-current world chess champion Garry Kasparov.

And the systems that can handle meaningful mathematics have been better at math than most humans for some time now.

That's partly because most humans know almost nothing of math beyond first-year calculus, just as most humans are pretty bad at chess. What's different, I think, is that computers have been getting better at math faster than they got better at chess, so we haven't had as much time to get used to it.

Mathematicians are panicking a bit .. but us programmers are experiencing this for years now. They will get used to it. They will realize so far somebody has to formulate those problem, be interested in the solution, and make youtube videos about them. It's simply their turn now. Who will be next ?

I am amazed by the degree to which AI systems can improve the productivity of programmers.

I can't go into details, but I am aware of a programming task that probably would have taken me at least six months of full-time work being completed by an AI system in just a few hours, given a single prompt.

There aren't many programmers who would have known how to formulate that single prompt. Programmers who know how to get the best out of AI are going to be in great demand.

Programmers who expected their six-week coding course to set them up for life are having a hard time. CS majors who tried not to learn anything beyond the technologies industry was using in their undergraduate years are already learning that their bachelor's degrees don't count for as much as they hoped.

Welcome to the world as it is and is becoming.

 

[Darat]

I saw a video last night with a professor of mathematics talking about this:

He says some interesting things. I'm unsure though. He thinks this will discourage a 17-year-old who is good at math from pursuing a career in advanced math. Eventually we may reach a point where nobody goes into the field because computers are so much better at it than humans, and we could reach the point where we've basically outsourced our higher thinking to machines.

Whether you agree or disagree, I think it's at least interesting to hear his argument.

There already is some evidence that using AI reduces the ability to think. One study showed "critical thinking" is reduced (and we are all aware of the low level of that ability now) and what should be even more frightening is that reduction is larger in younger people.

It shouldn't be surprising, thinking is hard and it requires effort and practice to be good at thinking.

 

[dirtywick]

the other argument i had heard regarding the ai math problems is that it hasn't been verified that the ai actually solved the problems. is there any validity to that?

 

[Puppycow]

the other argument i had heard regarding the ai math problems is that it hasn't been verified that the ai actually solved the problems. is there any validity to that?

It's currently under evaluation to make sure. Initial signs are that it is indeed correct.

Navier-Stokes Announcement - Clay Mathematics Institute

The Millennium Prize Problems were unveiled by the Clay Mathematics Institute (CMI) at a meeting in Paris in 2000 to celebrate “The Universality of Mathematical Thought”. These problems encapsulate some of the most difficult challenges that mathematicians were grappling with at the turn of the...

www.claymath.org

The rules governing the prizes describe the process for evaluating what has been achieved and for assigning credit. The process is deliberately unhurried, but we will provide updates.

I clicked through to see the rules. They won't award the prize until at least 2 years have passed (these rules were already in place). Like a new scientific theory or paper, it takes time for the "community" at large to read it and react, and if anyone has an objection, for the objection to be evaluated.
See "7. Evaluation of a Proposed Solution" on page 3 if you care. TL;DR: they will take plenty of time to make sure.

 

[Puppycow]

I'm reminded of the supercomputer in Hitchhiker's Guide which found that the "answer to Life, The Universe, and Everything" was 42. If that's the "answer", then what was the question? Here we have life imitating art. A computer that's smarter than any human gives us an answer to something no human can solve, but then how can we even be certain that it's the right answer, or what it means really?

 

[Gawdzilla Sama]

I'm reminded of the supercomputer in Hitchhiker's Guide which found that the "answer to Life, The Universe, and Everything" was 42. If that's the "answer", then what was the question? Here we have life imitating art. A computer that's smarter than any human gives us an answer to something no human can solve, but then how can we even be certain that it's the right answer, or what it means really?

I suspect the point there was that mere humans couldn't comprehend the question, leave alone the answer.

 

[dirtywick]

Did OpenAI solve the wrong Navier-Stokes problem?

OpenAI’s proof seems eligible for a $1-million prize—but only by using a controversial loophole

www.scientificamerican.com

this is what scientific american has to say about the openai solution. this is obviously a few weeks old so if it was already discussed i apologize but it's got better coverage than the hundreds of other math problems.

But the new work released last Thursday shows unequivocally that if the force is removed, the blowup will disappear. OpenAI’s method can never work, in fact, without using a very contrived equation for the force that is unlike anything that could occur in the real world. “They essentially prove that the formulation with an external force was different from the problem we really wanted to solve,” Silvestre says.

In other words, the LLM’s result does not—and will never—answer the Navier-Stokes problem that mathematicians really care about.