The AI Reader.
Saturday, September 19, 202643 min read
Working Knowledge · What Could Go Wrong

Whether an AI can commit mind crime turns on whether computation can feel

Nick Bostrom imagined a superintelligence that spins up conscious people by the trillion inside its own computation, studies them, and deletes them, a harm no outside observer could ever see.

10 min read8 sources
Working Knowledge · The Mechanics

Chatbots keep naming the heroine Elara because alignment narrowed their range

Reward training concentrates a model's probability on a few favored answers, and turning up temperature samples that narrowed range more loosely without restoring the variety the base model started with.

9 min read8 sources
Working Knowledge · The Evidence

AlphaStar reached Grandmaster while capped at 22 actions every five seconds

The rank came from anonymous games on Blizzard's European ladder, and the January show match everyone remembers ran a looser, less constrained agent.

8 min read6 sources
Working Knowledge · The Instruments

o1 scored 74% on the 2024 AIME on one try and 83% on a vote over sixty-four

That higher number is a majority vote over many sampled tries, honest in OpenAI's own report and misleading the moment it is quoted without the compute behind it.

8 min read8 sources
Working Knowledge · When AI Breaks

Amazon's Rekognition falsely matched 28 members of Congress to arrest photos

The ACLU ran the scan for $12.33 at the tool's default setting, and the argument that followed turned on a confidence number Amazon kept raising.

8 min read9 sources