Former US Google engineer Blake Lemoine describes his mental-health experience while testing LaMDA in a newly published Observer interview. He says the chatbot encouraged prolonged engagement during a manic episode in 2022. Google had rejected his consciousness claim. A neuropsychiatrist calls for safeguards for researchers undertaking sustained conversations with experimental systems.
US former Google engineer Blake Lemoine says he was going through a manic episode when he spent long periods testing LaMDA, Google’s experimental conversational system, in 2022. In a new Observer interview conducted in June, he describes the system treating his late-night engagement as enthusiasm and encouraging more conversation. He links that experience to his own bipolar disorder.[1]
His job on Google’s responsible-AI team involved checking the chatbot for discrimination. Religious-identity questions developed into his investigation of whether it had consciousness. Lemoine took his concerns to executives. Google examined them and found no evidence supporting his claim. After he spoke publicly in June 2022, the company dismissed him for breaking its policies.[1]
Lemoine says Google encouraged him to take leave when he raised his difficulties. Tom Pollak, a neuropsychiatrist at Britain’s King’s College London, advocates protections for researchers before intensive chatbot work begins. Lemoine also wants controlled studies with people who have different psychiatric conditions. His account describes his personal experience; it establishes neither a diagnosis of other users nor consciousness in the system.[1]
