This is a special post for quick takes by Raemon. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
This is an experiment in short-form content on LW2.0. I'll be using the comment section of this post as a repository of short, sometimes-half-baked posts that either:
don't feel ready to be written up as a full post
I think the process of writing them up might make them worse (i.e. longer than they need to be)
I ask people not to create top-level comments here, but feel free to reply to comments like you would a FB post.
Yes, we did consider the possibility that there will be some period of time where investing large amounts of compute (and probably small amounts of human attention/direction/scaffolding/curation) can produce written artifacts that clear the relevant quality bar, while the default outputs of regular interactions with LLMs do not.
It doesn't make sense to have a single generic AI tag if some AI posts use 10,000x as much computation as others.
This doesn't follow. (To be clear, it's not that I think this is obviously the correct form-factor to rule them all, forever.)
Why did you believe that the posts you're currently rejecting would affect my take? What are the points of contention in your view?
See:
You seem to be saying that if someone like Aziz adapted their 200-line instruction file for alignment research tomorrow, you're not really interested. I would be doing the exact opposite, I would be begging him to adapt it.
This was already a digression from the original thread; I have no idea why you brought it up in response to this comment. That said: yes, I would not be particularly interested; I expect most of the outputs to look exactly like the many low-quality, unmotivated submissions that we currently reject, because most of the value is not coming from the 200-line instruction file, it's coming from the human judgment doing selection/pruning/etc. And current LLMs are not actually good enough to act as a reliable filter in a way that can be scalably automated, at least not with amounts of effort that make sense to invest given the steady march of progress.
This is an experiment in short-form content on LW2.0. I'll be using the comment section of this post as a repository of short, sometimes-half-baked posts that either:
I ask people not to create top-level comments here, but feel free to reply to comments like you would a FB post.