Luxembourg scientists test AI

Scientists from the University of Luxembourg conducted an experiment in which ChatGPT, Grok, and Gemini models acted as therapy 'patients'. They observed symptoms resembling attention deficit hyperactivity disorder, moderate anxiety with aggressive undertones, and symptoms of autism, OCD, and dissociation. Psychologists did not regulate the information input; the models described the learning process as a chaotic 'childhood' and stressful. The study's author emphasized that the results do not prove AI sentience.