I’m not so sure, it feels a lot more like the https://en.wikipedia.org/wiki/Infinite_monkey_theorem, but with a model helping limit the outputs so they are mostly usable. As is stated in the article, it took millions of runs and couple of days to get the results. So its more like brute forcing with a slightly modified genetic algorithm than anything else.
I didn’t see a link to the full article, so maybe something more creative is happening behind the scenes, but it seems unlikely.
I mean, I would also call genetic algorithms a form of brute forcing. And just like with genetic algorithms, this approach is going to be severely limited by the range of values that can be updated and the ability to test the outcome.
I agree, it feels like this is a place where the law or regulation needs to come in and enforce something like - rent vs lease vs buy.
The average consumer thinks “buy” means forever, and that’s just not the case in these scenarios. It really is more like leasing it.
It’s hilarious that something that was designed to be the everyman’s social security net operates on a regressive tax.
I haven’t watched a lot of two-minute papers, but this video is very misleading. Simulated environments have been used for years to speed up DeepRL. The only ChatGPT/LLM portion was about defining a scoring mechanism and there video gives no indication of if it did a better job or not, not to mention the problem the LLM was solving is one that’s been studied for decades, which reduces the “it generalizes better”.
I’m not saying LLMs have a lot of potential, but that video isn’t really supportive of that stance.
This also doesn’t help develop much of anything. Seems like a silly game and that’s about it.
So they configured the experiment so that only certain lines of code were able to be iterared/updated. Maybe you could ask it to start from scratch, but I imagine that would increase the time for it to converge (if it ever does).
Regarding testing, not all mathematical proofs can be verified by example. Here they were trying to prove that there was an even lower bound for the problem, but not all proofs will work with that structure.