Google, a subsidiary of Alphabet Inc., announced on Friday that it had fired a senior software developer who had asserted that LaMDA, the company’s AI chatbot, had a sense of self.
Blake Lemoine, a software engineer, was put on leave last month by Google, which stated he had broken company rules and that it had found his assertions on LaMDA to be “wholly baseless.”Blake continued to consistently breach explicit employment and data security regulations, including the requirement to protect product information, despite extended discussion on the subject, a Google representative said in an email.
Google said last year that LaMDA, or Language Model for Dialogue Applications, was based on its research demonstrating that Transformer-based language models trained on dialogue could pick up pretty much any language.
Lemoine’s ideas were swiftly dismissed as misguided by Google and many other top scientists, who claimed that LaMDA is just a sophisticated computer made to produce believable human language.