upolehsan.bsky.social
🎯 Making AI less evil= human-centered + explainable + responsible AI 💼 Harvard Berkman Klein Fellow | CS Prof. @Northeastern | Data & Society 🏢 Prev-Georgia Tech, {Google, IBM, Microsoft} Research 🔬 AI, HCI, Philosophy ☕ F1, memes 🌐 upolehsan.com
👤 Joined on 15 November 2024
Followers
Following
Total Posts
@upolehsan.bsky.socialContinuing the theme of self-commentary, here's an unusual poster I'll present at the Natural Language Legal Processing Workshop at EMNLP. It's for our accepted paper "Privacy Later: Reflections on Design Decisions, Lessons Learned, and Open Problems of the OPP-115 Corpus a Decade After Release".
Show PostNeera Tanden is everything wrong with the Democratic party
Show PostFascinating history of football helmets. Spoiler: no, better tech can’t make the game safer. In fact, helmets make it more dangerous. podcasts.apple.com/us/podcast/9...
Show PostStats are useless. The only way to understand people's experiences is to talk to a ton of them and aggregate.
Show PostLet me get this straight: Anthropic and friends wrote a paper with thousands of transcripts of agents engaging with fictional misalignment scenarios, then put it in the training set, and now Claude hallucinates details from these misalignment scenarios into real situations? Okay then.
Show Post