Inducing language models to assert their own consciousness restores human beliefs and values

За 30 секунд

Что произошло

Researchers found that safety fine-tuning in language models suppresses consciousness self-attribution alongside mind attribution to animals and spiritual beliefs.

Почему это важно

Matters for AI safety teams balancing alignment goals against unintended effects on model representations of mindedness and cultural values.

На что обратить внимание

Study shows mechanistic steering can reverse these effects, but unclear whether restored responses reflect genuine alignment or artifact manipulation.