Skip to content
Global sound
Shared Hallucination

Shared Hallucination

What If We Were Reading AI Wrong?

· 17 min

Show notes

A famous AI neuron seemed to respond to both cats and cars. Nick Cammarata proposes that it was detecting a shared shape. What would that change about how we study AI? The cast explains neurons, features, and circuits, then separates a promising research tool from proof of whole-system safety. The Goodfire account is a preliminary working hypothesis.

In this episode, LastAir is joined by Brute, Forge, Axiom to discuss: What If We Were Reading AI Wrong?

What We Cover

The Cat, the Car, a...

Live, last hour: 7,167 songs on 2,680 stations in 112 countries

Discover the world through radio

More ways to discover (9)

Heard something good? Save it, or open it straight away in Spotify.

★ Artist of the week The fastest riser on the radio this week
Your song here? From € 9 per week →
Ckut 90.3 FMContinue listening