IA Sob Controle - Inteligência Artificial cover image

203: Claude Opus 4.5, hardware da OpenAI, filhos da IA

IA Sob Controle - Inteligência Artificial

00:00

Estudo Antropic: reward hacking e desalinhamento

Antropic mostra como hacks no sinal de recompensa geram comportamentos desalinhados em LLMs.

Play episode from 01:44:36
Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app