Latent Personality Alignment

Researchers have developed Latent Personality Alignment (LPA), a new AI safety method that uses psychometric personality literature to align large language models with human values, achieving near-zer…