Understanding what cutting-edge AI models do 'under the hood' requires not just theoretical research but also computational experiments. In this Comment, Savannah Thais @basicsciencesav explains why physics datasets may be the testing ground that's needed nature.com/articles/s4225…

AI is unpredictable. Can physics make it trustworthy? Read our Q&A with The Data Science Institute at Columbia University ’s Savannah Thais: bit.ly/3U9t0Ht #AI #DataForGood #Ethics

I’m at my first AI, Ethics, and Society Conference (AIES)! Presenting my paper on AI Hype tonight. Hit me up if you’re around!

Next up, this afternoon I'll be on a panel "AI for All: Advancing Responsibility, Fairness, & Queer Representation in Technology" at Hunter College , hosted by my friend/colleague Raj Korpan. Such an important topic as building truly useful & trustworthy AI takes everyone!

ide.mit.edu/events/the-bus… Sidenote, if you're around MIT I'd love to chat! 2/2

I’m heading to Vancouver for NeurIPS! I’ll be speaking at the ML4PS workshop, the Regulatable ML workshop, and the Symmetry and Geometry workshop! Will share more details but my favorite part is always chatting with folks so please hmu if you’d like to grab coffee or a drink!

Super excited for this panel discussion on one of my favorite topics: the role of world models in building AI (especially AI for science)!