I enjoy Jerry Fodor’s framing of this point (web.ics.purdue.edu/~drkelly/…, excerpt below); @davidchalmers42 has also said for awhile that any solution to consciousness will involve at least one crazy-sounding idea, which I agree with. I’m sure there are a few crazy-sounding proposals I left out

Only a tiny fraction of the world’s mosquito species prey on humans. All of these are invasive across essentially their entire range, having tagged along with humans as we moved around the world. Multiple careful studies say that eliminating these species would have little or no impact on the environment. However, it would save at least a million human lives a year, and get rid of a constant source of discomfort that prevents people from safely spending as much time outside as they would prefer. Malaria is only one of dozens of diseases, many potentially fatal, spread by mosquitoes. Gene drives can safely and relatively inexpensively eliminate these parasitic insects; we can create male mosquitoes that can only have male offspring and which then crash the population. It’s well past time that we did it.

My personal favorite Muse gadget hack so far - an ambient eInk display that sits above my keyboard where my @Muse pushes useful, fun or interesting content throughout the day.

these really are the geocities days of AI, we should remember to enjoy them before everything gets too slick

New paper on the mech interp of sycophancy! We find a small set of mid-layer attention heads that encode the user's beliefs. Ablating them eliminates most sycophantic agreement! Check out the preprint👇

If you think that AI—even if it works safely and fabulously—will make future human pursuits ‘pointless’, you don’t have a very pro-human anthropology

🚨New Preprint alert: Introducing CogGym: a benchmark for measuring human and Ai alignment using 280+ computational cognitive science experiments sourced from 100 studies. TLDR; they are far from being aligned. What is cool is that it leverages a semi-automatic pipeline for converting experiments into common format that is runnable on machines and humans (think replicability), making it a living and scalable benchmark. Importantly, it is also is a step toward expanding experimental design space for automated cognitive science. Stay tuned for that :) Congratulations to @LanceYing42 and team for this monumental effort!! Check out the thread ⬇️

I stand by this line - interpretability MIGHT be a big deal, and is certainly good enough to be useful, but is nowhere near the level of quality and reliability where anyone should be relying on us to ensure things go well

Future generations won't believe we let this happen.

Students sometimes ask me if it still makes sense, in this accelerating age, to pursue a PhD in AI. Perhaps counterintuitively, I think it's a great time to do so. I wrote up some thoughts on this here: web.mit.edu/phillipi/www/wri…

Astronomers why have you been hiding this from us

One of my hot takes is that the nature of consciousness is a deep and difficult problem that has puzzled philosophers for centuries and is relatively unlikely to be resolved via Twitter debates.

Hello everyone! I'm organizing a short but intensive 1 month virtual seminar in October targeted to LatinAmerica undergraduates to help them prepare for their PhD/Masters applications in STEM fields (limited availability)! Please feel free to share🫡 luma.com/bqdqfpoh

Progress in machine learning is bottlenecked by having good evals, interpretability is no exception. We've made WorkspaceBench: a range of scenarios where we know what the model should be thinking about, to see if your interp tool can find it J-Lens is great but only outputs a single token. I think a good multi-token J-Lens should do well on WorkspaceBench!

Just catching up with how much better voice-to-text transcription has gotten and the fact that it can detect semicolons (?!) in speech somehow

The recent breakthrough in Navier-Stokes has garnered a lot of attention to the new possibilities that emerge when agents work together. We have been interested in this question for a while. How can a team of agents achieve more than agents working alone? I'm excited to finally share our new work, "Self-Organizing Agent Teams Learn to Reason Together" 🧵

thought this was another “non-negotiables for dating me” list for a second

This paragraph by Richard Feynman hits so hard: “Fall in love with some activity, and do it! Nobody ever figures out what life is all about, and it doesn’t matter. Explore the world. Nearly everything is really interesting if you go into it deeply enough. Work as hard and as much as you want to on the things you like to do the best. Don’t think about what you want to be, but what you want to do. Keep up some kind of a minimum with other things so that society doesn’t stop you from doing anything at all.”