A “read, but not fully understood” column.

A “read, but not fully understood” column. A bunch of stuff has piled up.
1/
a super damn enjoyable series of articles (https://lilianweng.github.io/)
about AI, I like the delivery and the content, it’s pretty technically dense. Worth noting the author is a cofounder of Thinking Machines and before that VP AI Safety etc and R&D at OpenAI
2/
Storonsky’s playbook (https://quantumlightcapital.com/playbooks/performance-management/talent-philosophy)
also been lying around. No idea how to apply it at a team-of-a-couple-dozen level. But if it’s suddenly useful to someone.
3/ everyone who’s into A/B tests and the whole “one thing is the cause of another” rather than just correlation
I recommend reading (https://alexdeng.github.io/causal/intro.html)
4/ found out what an
analemma is (https://www.astronet.ru/db/msg/1213202)
5/ an opinion is generally starting to circulate already, (https://blog.jxmo.io/p/there-are-no-new-ideas-in-ai-only)
that with the current generation of nets we won’t see a breakthrough, we’ll only see the results of better-labeled data
6/
Feynman (http://genius.cat-v.org/richard-feynman/writtings/letters/problems)
“
No problem is too small or too trivial if we can really do something about it.”
7
/ Read a couple chapters of Schopenhauer. Realized that philosophy is of course cool, but the practical value isn’t that big. “The World as Will and Representation”.