OpenAI Researcher Suggests Brain Sensors for Emotional AI

Gabriel Petersson, a research engineer at OpenAI, posted on X suggesting reinforcement learning with brain sensors could teach models to recognize tokens that evoke emotions such as awe, disgust, or happiness. He stated that AI lacks the ability to know what emotion it evokes in people. Petersson gave the example that current systems cannot write beautifully because they have no sense of when or how readers react. The post presents the idea as a question rather than a completed experiment.