Join us on Wednesday April 23 at 12 PM ET for Metagov Seminar: A rule-based approach to mitigating AI-risk (Evan Miyazono). For more details and link to join, register here - lu.ma/m0lc3280

Attending #ICSE2025 in Ottawa? Join us at the Rust Symposium (Friday, May 2, 9am-5pm) for a full day of cutting-edge tools and research. Featuring speakers from AWS, ETH Zurich, Formal Land, Inria, Microsoft, and Runtime Verification. #RustLang Registration here lu.ma/umi3g2wc

I was recently asked "has there ever been a technology (other than weapons) where there was as much debate about whether it would be good for society as there is for AI?" I'm unusually pleased with my answer: "the creation of the US government"

I want to be able to mash on my keyboard to select the next most likely autocomplete/autocorrect option. I think I'd tolerate much more aggressive autocorrect if I had this affordance. The voice command "ugh" would also be sufficient.

It's easy to find opinions+evidence that "o3 is a superhuman problem solver." Also easy to find takes+evidence that it's way dumber than a child. The surprising takes are the ones that assume only one of these is true.

I have a solution I love for the remote-first company holiday gift exchange. A recommendation exchange! Everyone gives and gets a personalized recommendation, no shipping involved - here's how we did it:

"a comprehensive and incomprehensible deep dive" *chef kiss* (stolen from youtu.be/qwiLnLsuqx8?...)

The Goodhart-Hanlon lazor: Any sufficiently large group or entity appearing to act maliciously likely includes, somewhere, at least one unaware and/or unwilling person nonetheless making things worse by just following orders and maximizing some metric.

I knew FutureHouse generated a mind-numbing number of Wikipedia pages, but I was very surprised to learn that human peer review found 2x the errors in human-generated articles compared to the AI-generated ones

I love that Cosmos Ventures includes "supporting individual preferences and abilities" in its definition of human flourishing. "Would a different person make the same choice" seems to be a really interesting indicator of freedom/agency.

"generative adversarial proving" One agent tries to prove the theorem; one tries to find counterexamples

I realized I support people because I believe in them - I want to amplify/empower them as an embodiment of their preferences and values. Not because I like the values themselves.

I'm combating fomo about not being added to starter packs with the belief that it means I'm just part of some "advanced" tier for connoisseurs

I don't.... Feel like.... We're grappling with how beyond nuts this is.... I do not have children nor do I intend to but I care about human society and being part of it and therefore childcare feels like one of THE issues we absolutely must solve as a nation, how does everyone not think this

TIL: phalanges (aka finger/toe bones) is a plural of phalanx (ancient Greek military formation)

Bringing Military x AI posts from other socials: I cannot emphasize enough how much of a disaster this portion of Meta's announcement is for safety engineering and will guarantee a catastrophic incident. TechCrunch covered our criticism here: techcrunch.com/2024/11/04/m...

> only 3 ways for a scientist to guarantee positive results: - Be a psychic; - Study only marginal, incremental topics where... you can virtually guarantee a positive result; and/or - Skew your research design, data, and analysis, and hide any results that are still null.

cold take: if your idea about the world doesn't include a testable mechanism, it's risky pseudoscience or philosophy spicy take: that includes your favorite interpretation of quantum mechanics and any genetic factors for intelligence

I really want "spec-based AI" to be a term: specification (aka spec) - a set of objective criteria any solution should be verifiable against. spec-based AI - AI systems where outputs can be guaranteed against those objective criteria

Open protocols are the new antitrust because network effects are the new vertical integration