Employees of the USnn food and medicine control department told CNN that the new AI system called the ELSA, which was recently introduced by the department, sometimes invents non-existent scientific research or distorts real data. One of the employees noted: “If there is no time to double -check, you cannot trust. She is confidently “hallucinating”. ”
ELSA is designed to accelerate clinical examination and assistance in making decisions affecting patients’ health. However, such errors can be dangerous.
FDA management has not yet considered the problem of serious. The Commissioner of the Agency Marty Makarai told CNN that he had not heard specific complaints and recalled that the use of ELSA and training in working with it is still voluntary.
Information about the problems with AI appeared on the same day when the White House presented its “plan of action in the field of AI.” The document refers to the US desire to win in the technological race, simplifying the regulation of AI and the elimination of the so -called “ideological distortions” – for example, mentions of climate.
