I am delighted to share our new #PNAS paper, with Gaurav Kamath, Michelle Yang Siva Reddy and Morgan Sonderegger, looking at whether age matters for the adoption of new meanings. That is, as words change meaning, does the rate of adoption vary across generations? pnas.org/doi/10.1073/pn…

The idea that ‘historian’ is one of the jobs most at risk from AI is darkly funny; after all, in a world dominated by often-wrong AI what use could *a set of people with specialist training in chasing down and verifying the sources of information* possibly be?

Friend from ultimate frisbee was at cool ai startup -> joined part-time as undergrad and soaked up ambitiousness Always loved philosophy so started engaging with ai from that angle etc Was in comp linguistics BSc -> reached out to profs eg hinrich schuetze for research

New editing model released! It's been fun to learn about unified autoregressive models on this project We learned they're tricky to tame but I think ultimately they're the future of deep vision & language integration

We didn’t have 3,000 authors like GDM, but we collaborated at full force! Saba brought this tough project to life. Benno Krojer wrote my favorite intro ever. @aa_kazemnejad added the RL cherry on top — and it worked! Ankur Sikarwar crushed it from ideation to execution.

I've worked on image editing academically but it's cool to see a first real use case in my personal life with my mom: rearranging and adding furnitures from different houses; and it worked first try (with chatgpt)!

this isn’t just a modeling problem. it’s also a benchmarking problem. spurious correlations are always a pain, but in multimodal llms they become a particularly tough battle. On one hand, you want to leverage the language prior to enable better generalization; on the other, that

This is precisely why we released MVPBench to measure genuine video understanding with minimal video pairs Otherwise we will continue to overestimate physical/visual understanding in LLMs arxiv.org/abs/2506.09987

Can highly recommend making important slides while sleep deprived Your brain will be so slow and low attention span that you will keep slides much simpler and with less text, getting the core message across

Feeling confident to say that Conference on Language Modeling is the top AI/LM/ML conference this year as far as culinary location! Montréal ❤️

This is an awful vision of the future. Do not build this.

Appropriate beer design after passing my phd proposal exam

Congratulations Benno Krojer ! A great proposal and exciting work!

The German in me is really upset that the definition of brunch got so stretched that places offer brunch until 3pm these days

Excited to announce this awesome workshop we are organizing at Mila World models touch so many fields from robotics/videos to LLMs and AI4Science: there will be something for everyone! We already have several amazing speakers confirmed, with more to be announced soon 🤩