This is a special post for quick takes by Alexei. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
Pretty obvious, but: when applying the Chesterton's Fence principle to decide whether or not to remove the metaphorical fence, consider two things:
- Remember that the existence of the fence itself has a cost. If it's an object, it's cluttering up space and might demand repair / organization. If it's a rule, it's demanding attention and detracts from other rules. Etc...
- Removing the fence will likely remind you what it was for when eventually something breaks. Value of information!
I’d make it highly recommend / mandatory on LessWrong for all posts at the top to disclose to what extent they have been generated using ideas from AI. I think a rule like that is already in place if the post is partially AI written. And I think knowing where the ideas are coming from is just as important. I’m beginning to suspect that more and more posts (everywhere) are influenced in large part by AI. In which case, why am I even here?
I do really think there's a lot of using-AI and a dearth of understanding-how-people-are-using-AI.
I think using AI is often good, but I would like more understanding of what's going on.
There are the accounts Claude+ and chatgptopenai, which one could make co-authors on posts to indicate idea/feedback/whatever-involvement. In general, I'm happy to support developing norms about AI involvement, as to give people an out if they're worried about memetic plagues.