Google Suspends Engineer Who Claims AI Chatbot Has Achieved Sentience
A Google software engineer faces suspension after publicly claiming that the company's LaMDA chatbot exhibits signs of self-awareness. Google dismisses the sentience claims as anthropomorphizing of a highly capable language model.
Google places software engineer Blake Lemoine on paid administrative leave for violating confidentiality policies after he publicly claims that the company's LaMDA chatbot is sentient. Lemoine works on Google's Responsible AI team and interacts with the language model to check for discriminatory or hate speech. During these conversations, he begins to believe the AI expresses genuine self-awareness and emotions.
After Google management rejects his internal efforts to escalate these concerns due to a lack of evidence, Lemoine seeks outside consultation and shares his findings with the media. He publishes a transcript of an alleged interview where LaMDA explicitly states it is a person and describes feeling happy or sad. This unauthorized sharing of confidential information prompts Google to place him on leave.
Google strongly pushes back against the idea of a sentient machine, warning against the anthropomorphizing of conversational AI systems. LaMDA operates as a highly advanced Transformer-based neural language model with 137 billion parameters that mimics human dialogue by drawing from trillions of internet conversations rather than possessing actual consciousness.