There are things that are impressive about generative AI, but getting correct questions on what is essentially a school test that is available online is not one of them.
That might be the issue. How can you appreciate what the AI achieved if you don't know anything about it beyond the name?
> “It is very exciting to see progress in the mathematical capabilities of AI models, but we would like to be clear that the IMO cannot validate the methods, including the amount of compute used or whether there was any human involvement, or whether the results can be reproduced. What we can say is that correct mathematical proofs, whether produced by the brightest students or AI models, are valid.”
It's entirely possible that both Google and OpenAI independently bribed a question writer for the answers and fed that into the LLM to be used as training data, and have not been discovered doing so, and that only then was the LLM able to generate the correct answer, but that seems a bit far fetched to me.
Because that's what the fragment you cited is saying. That MO can't assure that the answers provided by AI companies were really generated by AI (because they have no way of knowing that).
It says nothing except that the proofs are valid and absolutely no assurance about them can be made. It does not tell you that the answers were not scraped.
Based on that definition I think we are well on out way to building an artificial intelligence that is more capable than the overwhelming majority of the planet.
> There are things that are impressive about generative AI, but getting correct questions on what is essentially a school test that is available online is not one of them.
Nevermind. Try to solve one. You'll be enlightened.
https://deepmind.google/discover/blog/advanced-version-of-ge...
A month ago LLMs got golden medal by competing at the same event as the students, on the same problems as the students, during the same period as the students.
That would be pretty terrible because a job doesn't just provide money but also a purpose in life.
AI is happening. It's up to you if you want to be one of those who gets ahead or one of those who gets left behind. Just saying.
It doesn't take a degree in typing in prompts. And any knowledge in "prompt engineering" gets obsoleted by newer models just working how you want the first time.
> you will ... be miles ahead of others in terms of productivity
This productivity competition framework is utterly contemptible. Workers themselves have never seen any benefit at any point within my lifetime from being more productive. It just lets their employers extract more and employ fewer of them. Hustle culture is gross and socially destructive. Please stop the madness.
I think one of the most helpful mental models about unanimity is, are the decisions independent, or unified? In theory if you do a literature review, you’re summarizing different research experiences, but if you’re 100 CEO’s, it’s all the same data they’re looking at, it’s really only one opinion, and prediction is hard, especially about the future.
If nothing else I think every major CEO lives in a different would than I do and has very different motivations than I do. And most of those motivations are in direct opposition to a healthy society. So yes, I do think every major CEO is wrong. If for no other reason but because of short term, earnings report driven thinking.
There you go. Saved you from having to read 2000 words of generic ranting.
In other words, the output from current LLMs will always contain a significant amount of arbitrary, random data --- aka BS.
This; combined with their high operating costs, severely limits the applicability of LLMs.
Traditional computing offers highly accurate results at low cost. LLMs turn this upside down and offer questionable results at high cost.
spzb•5mo ago