You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I asked ChatGPT 6.1 Sol “Try to resolve the 68 unknown (category, property)-pairs on https://catdat.app/missing ” and with a bit of encouragement it claims that all but 4 can be resolved. It would be nice to incorporate these properties into CatDat but the arguments would need to be checked/digested.
As for the pdf, I will need some time to read it, maybe even weeks. I hope you understand.
I also must say that it is a bit overwhelming to get all these proofs at once, especially because it takes a lot of time to digest 18 pages of AI-generated proofs (and there is always the danger that they are incorrect). I am not entirely happy about this. I am also not happy that an AI was fed with all the data in CatDat without my approval. Maybe I should have written some rules for this in the README, but in any case, it would have been nice to ask. Maybe I am also old-fashioned and this is just the way it is today; I cannot tell.
Of course, I will create multiple PRs for this (but others are invited too, of course!). Notice that there are also some isolated issues for deciding unknown properties (#317 and #91).
When I know more, I will write here. Till then, thank you again.
Ok, I probably should have asked, curiosity gets the better of me. Obviously there is no obligation for you to read this. I know other survey projects have struggled with AI contributions. The proofs I have looked at do look sensible.
Actually I left it running and it claims to have arguments for the remaining SchR properties except for effective cocongruences. The proof are very involved though; maybe I will at formalisation.
Activity
ScriptRaccoon commentedon Oct 7, 2026
Thank you for the initiative.
As for the pdf, I will need some time to read it, maybe even weeks. I hope you understand.
I also must say that it is a bit overwhelming to get all these proofs at once, especially because it takes a lot of time to digest 18 pages of AI-generated proofs (and there is always the danger that they are incorrect). I am not entirely happy about this. I am also not happy that an AI was fed with all the data in CatDat without my approval. Maybe I should have written some rules for this in the README, but in any case, it would have been nice to ask. Maybe I am also old-fashioned and this is just the way it is today; I cannot tell.
Of course, I will create multiple PRs for this (but others are invited too, of course!). Notice that there are also some isolated issues for deciding unknown properties (#317 and #91).
When I know more, I will write here. Till then, thank you again.
cmcq commentedon Oct 9, 2026
Ok, I probably should have asked, curiosity gets the better of me. Obviously there is no obligation for you to read this. I know other survey projects have struggled with AI contributions. The proofs I have looked at do look sensible.
Actually I left it running and it claims to have arguments for the remaining SchR properties except for effective cocongruences. The proof are very involved though; maybe I will at formalisation.
ScriptRaccoon commentedon Oct 9, 2026
This comment provides an overview of the work on this issue. It will be updated regularly.