Hi, so I've spent the past almost-decade studying research uses of public social media data, like e.g. ML researchers using content from Twitter, Reddit, and Mastodon. Anyway, buckle up this is about to be a VERY long thread with lots of thoughts and links to papers. 🧵

what you're afraid of: AI telling your microwave to kill you in your sleep what you should be afraid of: AI telling your insurance to reject your life saving surgery because it was trained on medical texts from 1853