metiulekm

@metiulekm@sh.itjust.works

• ⁨Comment⁩ on ⁨Are LLMs capable of writing *good* code?⁩ ⁨⁨2⁩ ⁨months⁩ ago⁩:

  Hasn’t Google already made advances through its Alpha Geometry AI?? Admittedly, that’s a geometry setting which may be easier to code than other parts of Math and there isn’t yet a clear indication AI will ever be able to reach a certain level of creativity that the human mind has, but at the same time it might get there by sheer volume of attempts.

  Wanted to focus a bit on this. The thing with AlphaGeometry and AlphaProof is that they really treat doing math as a game, not unlike chess. For example, AlphaGeometry has a basic set of rules, it can apply them and it knows when it is done. And when it is done, you can be 100% sure that the solution is correct, because the rules of the game are known; the 28/42 score reported in the article is really four perfect scores and three zeros. Those systems do use LLMs, but they really are only there to suggest to the system what to try doing next. There is a very enlightening picture in the AlphaGeometry paper here: www.nature.com/articles/s41586-023-06747-5#Fig1

  You can automatically verify correctness of code the same way. For example Lean, the language AlphaProof uses internally, can be used for general programming. In general, we call similar programming techniques formal methods. But most people don’t do this, since this is more time-consuming than normal programming, and in many cases we don’t even know how to define the goal of our code (how to define correct rendering in a game?). So this is only really done when the correctness of the program is critical, like famously they verified the code of the automatic metro in Paris this way. And so most people don’t try to make programming AI work this way.

• ⁨Comment⁩ on ⁨Dark Souls would’ve been a better game in early access, says dev behind Dark Souls-like early access game No Rest For The Wicked⁩ ⁨⁨6⁩ ⁨months⁩ ago⁩:

  I can personally say that I got super excited by the new release from the Ori devs at first, though later became disinterested because the game is so different. The Ori games weren’t obscure by any means, so I am not surprised other people got excited too.

• ⁨Comment⁩ on ⁨C and C++ Prioritize Performance over Correctness⁩ ⁨⁨1⁩ ⁨year⁩ ago⁩:

  Edit: Actually, I thought about it, and I don’t think clang’s behavior is wrong in the examples he cites. Basically, you’re using an uninitialized variable, and choosing to use compiler settings which make that legal, and the compiler is saying “Okay, you didn’t give me a value for this variable, so I’m just going to pick one that’s convenient for me and do my optimizations according to the value I picked.” Is that the best thing for it to do? Maybe not; it certainly violates the principle of least surprise. But, it’s hard for me to say it’s the compiler’s fault that you constructed a program that does something surprising when uninitialized variables you’re using happen to have certain values.

  You got it correct in this edit. But the important part is that gcc will also do this, and they both are kinda expected to do so. The article cites some standard committee discussions: somebody suggested ensuring that signed integer overflow in C++20 will not UB, and the committee decided against it. Also, somebody suggested not allowing to optimize out the infinite loops like 13 years ago, and then the committee decided that it should be allowed. Therefore, these optimisations are clearly seen as features.

  And these are not theoretical issues by any means, there has been this vulnerability in the kernel for instance: lwn.net/Articles/342330/ which happened because the compiler just removed a null pointer check.