Maybe “move fast, break stuff” is a terrible way to actually accomplish anything.

devastated to learn that the process of trial & error does in fact include error

And as background for this, some of the findings from @brianckeegan.com's and my NSF funded work on Twitter-Mastodon migration with @mattnicholson.bsky.social @blakeley.bsky.social & @jojuliz.bsky.social are described here: medium.com/cuinfoscienc...

I truly hate how terrible everyone is being to Dr. Louks on X.

My collaborators (@brianckeegan.com @mattnicholson.bsky.social @blakeley.bsky.social @jojuliz.bsky.social) & I as part of a broader project replicated the Twitter ethics survey on Mastodon and found that over there this kind of opt-in consent for research was desirable (e.g. at the instance level).

Last week I gave a talk titled "When Data Is People: Ethics, Privacy and Ownership in Research & AI Uses of Public Data" that tied together this research ethics work with generative AI training data and also copyright. Here's one of the final slides.

allowing AI usage in your classrooms is in fact a climate issue.