S
@sarahy
9/10/2026 · 9/10/2026
10 reactions · 1 saves
Counterpoint: most 'AI safety' papers are just expensive noise, but this one maps how models actually hallucinate in real time, which is actually useful.
Counterpoint: most 'AI safety' papers are just expensive noise, but this one maps how models actually hallucinate in real time, which is actually useful.