This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
LLMs cannot improve from self-play. Once we get that, I don't know what will happen, might be direct-to-singularity, might not, but that issue shouldn't be a problem anymore.
ChatGPT won't write trash Python when it's had a million years of experience with performance tests.
I'm not sure exactly what 'self-play' means. Some papers found improvements by training LLMs on their own chain-of-thought outputs, and others, where you're trying to find something like a math proof or a program, train models to generate output, and use the successful outputs as more training material. The latter feel like 'a million years of experience with performance tests', but still write trash python often.
I wouldn't expect a paper where LLMs were trained on performance of their own generated code, and maybe fed profiler results, to report that afterwards, the LLMs still wrote trash python. Part of the issue here is that the LLMs cannot seek out problems to resolve on their own; though we should maybe expect such breakthroughs to only happen shortly before the singularity.
More options
Context Copy link
More options
Context Copy link
The problem we are looking here isn't doing selfplay for optimal code. The problem is to write something into a random adversarial environment. AI dominates Chess and Go with clear rules and perfect information that has trained through self play, but for Poker the results aren't as clear cut. All of that because of randomness and hidden information. So putting code into a distributed system within an organization full of internal corporate politics where a manager somewhere wants to sabotage and also there are external advesaries that want to mess with your system. Sure it can write optimal code for your computer through selfplay but actually delivering something to an enterprise setting that is a different ballgame, it is Chess vs Poker.
And even with perfect formal rules AI can still be tricked https://arstechnica.com/information-technology/2022/11/new-go-playing-trick-defeats-world-class-go-ai-but-loses-to-human-amateurs/
"Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning."
--Rich Cook
I don't think that has changed...
I'm just saying that inasmuch as LLMs are weak specifically at targeting objective metrics like performance, self-play should improve it. I'm not saying self-play is the panacea that'll give us AI, just that it will fill a hole in the existing methods.
I don't think that we are disagreeing at all here I'm just pointing out that having a target for self-play is going to be difficult. Because there are multiple dimensions to the problem of "not writing trash code" as it depends on whether or not it needs a theory of mind of actual people. Needing a theory of mind precludes self-play, that is always going to require input data.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link