Reads.
A short list of what shaped how I think about AI.
Books
-
If Anyone Builds It, Everyone Dies
The strongest version of the loss-of-control argument. You don't have to agree with every claim to think it's worth understanding.
-
The Precipice
The rigorous accounting of existential risks, natural, anthropogenic, and technological, and what to do about each.
-
What We Owe the Future
The case for long-termism.
-
Superintelligence
Certain parts have aged. The core arguments haven't.
-
The Alignment Problem
A journalist's tour of the field. The best entry point for someone who has never read anything about alignment before.
Articles
-
AI 2027
A detailed scenario of how the next few years could play out. Not a prediction. An existence proof that the path from here to loss-of-control is short.
-
Rationality: A-Z
The sequences are tight enough to read in minutes, which makes an individual sequence the wrong recommendation. The whole thing is the recommendation. Long, but worth ripping through.
-
The case for AI control
The case for control as the short-term solution.
Channels
Not exactly reads, but worth your time anyway.
-
Rob Miles
One of the reasons a lot of people in my generation are working on alignment. Amazing tone and content.
-
AI in Context
I've personally met Aric, and wholeheartedly recommend his content.
-
80,000 Hours
For career guidance. Their podcast is great too.