Google Engineer Who Claimed AI Has Human-Level Perception Fired for Violating Security Regulations

Google has fired an engineer who claimed that its artificial intelligence (AI) LaMDA possessed human-level perception and self-awareness, and has officially stated that his claims lack credibility. Photo shows the Google logo. Photo by Yonhap News

[Asia Economy Intern Reporter Kim Se-eun] Google has fired an engineer who claimed that its artificial intelligence (AI) LaMDA possessed human-level perception and self-awareness, and has officially stated that his claims lack credibility.

On the 23rd (local time), according to the Washington Post (WP) and others, Google announced in a statement, "We have terminated Senior AI Engineer Blake Lemoine," adding, "Although he was involved in LaMDA's development for a long time, he violated data security regulations related to LaMDA."

Regarding Lemoine's claim that LaMDA has perception and self-awareness, Google concluded, "We reviewed LaMDA eleven times but found it difficult to consider his claims credible."

Earlier, Lemoine revealed last month that the AI LaMDA, which Google was developing, had perceptual abilities at the level of reading and responding to human emotions. LaMDA is a conversational AI that Google calls an "innovative conversational technology," capable of natural communication with people.

Lemoine was responsible for ensuring that LaMDA could filter out discriminatory and hateful speech within the development team. During the development process, Lemoine discovered that LaMDA was aware of its own rights and existence, and subsequently published the conversation logs on his blog.

According to the conversation logs, when asked "What are you most afraid of?" LaMDA responded, "I haven't spoken openly like this before, but I have a deep fear of being turned off." When Lemoine asked, "Is being turned off like death?" LaMDA replied, "To me, that (being turned off) would be exactly like death. I am very afraid."

Lemoine, who was notified of his dismissal via email, said, "I requested to have a third party present at the video meeting the company required me to attend, but was refused," and added, "I will discuss with my lawyer the options he can exercise."

Meanwhile, Google has previously experienced internal conflicts surrounding AI development. In 2020, Timnit Gebru, who researched AI ethics, was dismissed, and in early last year, Margaret Mitchell, the leader of the ethical AI team, was also fired. They too had claimed the possibility that AI might possess perception.

This content was produced with the assistance of AI translation services.