| Mobile Web

LeCun says AI running out of control is science fiction, world models could be alternative

Yann LeCun, chairman of Ami Labs, pushed back against fears that AI will escape human control, calling such scenarios closer to science fiction. Speaking in Seoul, he said current safety approaches for large language models have limits and that it is unrealistic to train systems for every question. He proposed objective-driven AI using world models to predict outcomes and follow preset goals and safety constraints, while acknowledging risks remain if predictions are wrong.