I can't believe we allow for permissionless access to strong search engines. I think we should only allow for open weak search engines. If you do build a strong search engine, it's more responsible to destroy it honestly. What if people can search about bioweapons?

Congrats Elan Rosenfeld on a great thesis that moves our understanding of distribution shift forward! w/ Andrej Risteski Boaz Barak Uri Shalit

Congrats Bingbin Liu (and to Kempner Institute at Harvard University!)

Thrilled to announce the 2024 recipients of #KempnerInstitute Research Fellowships: Thomas Fel, Mikail Khona, Bingbin Liu, Isabel Papadimitriou, Noor Sajid, & Aaron Walsman! bit.ly/4aBQ6MS

Thomas Fel mikail khona Bingbin Liu Isabel Papadimitriou Noor Sajid Aaron Walsman

the premature gray hair, the smile without the smile, a small price to pay really

AI should be aligned around the values of Noah Smith. In other words, it should focus all its efforts on protecting rabbits and promoting anime.

On Friday, I successfully defended my PhD thesis 'Identifiable Causal Representation Learning: Unsupervised, Multi-View, and Multi-Environment' in Cambridge, examined by Profs Jose Miguel Hernández-Lobato and Pradeep Ravikumar. I'm particularly delighted to have passed with no corrections 🎉

CARE talks kick off again this week, with
Goutham Rajendran talking about learning disentangled representations (portal.valencelabs.com/events/post/le…). It's a really nice paper showing with linear + Gaussian latents we don't need many interventions to disentangle latents. Thursday, 11am EST

When you talk to young folks, they think that the belief in imminent AGI was caused by the rise of LLMs. In reality, this belief is axiomatic and long predates LLMs. DeepMind, Vicarious were founded in 2010 based on this belief. OpenAI in 2015. Interest in LLMs only began in…

I wonder what senior researchers think about the similarities between investments in (very performant) expert systems in the 80s and the investments in (very performant) large neural models now in the 2020s? Yann LeCun Thomas G. Dietterich?

It's Exponential Age Week! 🚀

Today, we have 2 great convos lined up for you:

1) AI deep dive with Pradeep Ravikumar, professor of machine learning & computer science at Carnegie Mellon University & David Mattin

2) Benjamin Miller, co-founder and CEO of Fundrise joins Maggie Lake

And then we get Duran Duran, and U2, possibly correlation rather than causation, but it all comes crashing down. Because the large-scale systems, called expert (rule-based) systems, in the 1980s, were very good but not good enough for enterprise use cases. 2/2

It's gratifying to see a lot of investment into large scale AI systems that seem to be the future of AI. Based on technology that is a smart ongoing refinement of something developed many decades back. And all this with the backdrop of inflation finally coming back down. 1/2

On twitter before the AI winter; the whitewalkers (the AI doomers) seem to be multiplying, with their ice dragons