How do machine learning models interpret paintings and optical illusions? Check out these blog posts I worked on with Emily Reif and Been Kim on depth in art history (pair-code.github.io/depth-maps-art…) and depth model illusions (pair-code.github.io/depth-maps-art…).

The availability of giant datasets and faster computers is making it harder to collect and study private information without inadvertently violating people’s privacy. Ellen Jiang and I took a look at how random numbers can help. pair.withgoogle.com/explorables/an…

Why Some Models Leak Data: Ellen Jiang and I visualized soccer and salary models to show how training data can be inadvertently revealed. pair.withgoogle.com/explorables/da…

Search results that reflect historic inequities can amplify stereotypes and perpetuate under-representation. Carefully measuring diversity can help. w/ Ellen Jiang dylan MMitchell @[email protected] on Mastodon explorable: pair.withgoogle.com/explorables/me… colab: colab.research.google.com/github/PAIR-co…

Do Machine Learning Models Memorize or Generalize? pair.withgoogle.com/explorables/gr… An interactive introduction to grokking and mechanistic interpretability w/ Asma Ghandeharioun, nada hussein, Nithum, Martin Wattenberg and iislucas (Lucas Dixon)

Belated update: We recently launched AutoNotes, an AI-notetaking experiment with open-source code + web demo now available! viv A. Connors Alejandra Molina Aaron Donsbach medium.com/people-ai-rese…