Curious how this experiment fares by the criteria laid out here: users.eecs.northwestern.edu/~jhullman/in...

What matters is the relationship between your claims, evidence and inference. You probably can’t do this with p<.05, but you certainly can by holding a fernandina tortoise.

I *literally* wrote a book about all this, and I'm just so tired. We've been trying to get people to pay attention for over ten years.

Led to them setting policy to cap submissions to… 6? Per person? And double down on prohibiting AI use in grant writing. This is all from my memory of a conversation, so cross-check as you will, but checks out and makes me … really angry.

Uggghhhhhhh

Also just heard today abt a single NIH study section getting flooded with 91 AI-generated grant proposals from a single PI

This looks really useful!!

Seems to me it matters that we stop trying to unify science around shared immediate objectives beyond basic norms (such as honesty and integrity), and have communities locally set their priorities and actively work toward them. Such pluralism will provide fertile grounds for learning and progress.

Call for papers and position statements is here, and closing this Friday July 11: human-ai-sensemaking.github.io

Please consider submitting, and share widely with your networks!

At its best, this work can connect still with social-science critique in a "build from the critique" arc/cycle.

I am trying to work on this now with critiques of infrastructure reform from a social science of infrastructure perspective, and prospectively designing/building from those critiques.