A theorem that was storied in my graduate studies--the Unique Games Conjecture is true! So many other results follow (under P != NP):
-> Basic SDP + Raghavendra rounding is optimal for approximating every Max-CSP
-> Trivial 2-approximation of vertex cover is optimal
-> Goemans-Williamson 0.878 approximation of MAX-CUT is optimal
-> Min-Uncut, Multicut, general weighted correlation clustering have no constant factor approximation
-> LLZ 0.940 algorithm for MAX-2SAT is optimal
@BooleanAnalysis has to revise his book!
https://t.co/N6kqrY5TOw
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
https://t.co/7N6TPlft1P
Zain epitomizes the brightest of a new generation of young tech founders—he is sharp yet warm, ambitious yet humble, and mature yet idealistic enough to envision a radically better future. I've personally never seen anyone learn and execute faster than him, and this is precisely why I'd never ever bet against him.
Congrats on the launch!
The top software engineers know intuitively that every software system must be refactored every time it grows roughly ten times in scale. It seems that a similar principle applies to most systems: great startups tend to and have to reorganize their structure as they scale; great managers start from managing a single task and qualitatively grow their skill to manage hundred of tasks by rewiring how they think.
Is AI actually speeding up drug discovery?
According to new McKinsey research, there are early signs that it is: all of these candidates were AI-enabled, with discovery time cut by ~15-80%.
(Incredibly bullish)
I see two ways in which this will turn out: the academy will either turn out batches of technically capable leaders or tech nerds with an narrow view lacking basic appreciation of the humanities and sciences who cannot do great work due to it. I deeply believe liberal arts education to some extent is necessary even for builders, but maybe for the top-tier talent such education up to high school might suffice.
That said I'm glad to see the disruption to education, as higher ed generally seems to fail in cultivating people who can actually do great work
(also check out this article from today's wsj: https://t.co/0knX32nANj)
pro tip for language learners:
ask Codex to connect your chatbot with your Mac’s Dictionary using MCP! and then ask them to cite dictionary entries as evidence for comprehension, understanding concepts etc
100% spot-on, this is a must read
what we need is a way for humans to appreciate machine- or human-generated proofs more efficiently, not "alignment" with humanity and slowing down mathematical progress.
were mathematicians ever capable of digesting all human-written proofs anyways?
https://t.co/Orb5eHVboe
SITUATION DETECTED: Anthropic has set up a wet lab in the San Francisco Bay Area for physical biology work. It wants to unlock treatments for rare diseases, and its research has now gone beyond in silico evaluations, per Reuters.