Google suspends engineer who claims LaMDA has become self-aware

Google has laid off Blake Lemoine of the company’s Responsible AI division. Lemoine made headlines over the weekend over claims that the LaMDA conversation model has taken on a life of its own and has a soul.

Google has suspended the developer for violating the company’s confidentiality policy with its publications. Among other things, Lemoine published logs of chats he had with LaMDA. In addition, he had gone to The Washington Post with his claim that LaMDA had become self-conscious.

The developer, together with another employee, had come to that conclusion based on conversations with the language model. They tried internally to convince the company’s vice president and the head of the Responsible Innovation department, but they denied their claims.

Google unveiled its Language Model for Dialogue Applications, or LaMDA, at its I/O 2021 conference. The language model makes it possible to have fluent conversations on many topics, similar to how people converse with each other online. LaMDA is trained on large amounts of data for such back-and-forth conversations.

At Google I/O 2022, Google announced that LaMDA would become part of its AI Test Kitchen, allowing a limited group of test users to try it out. Google then emphasized that it wants to use the technology ‘responsibly’ and use user feedback to improve the language model.

Blake Lemoine published an article over the weekend in which he claimed, based on his conversations, that LaMDA “is consistent about what it wants and what its rights are as a person† LaMDA would like Google to request permission for experiments and the model would like to be recognized as an employee of Google. He wondered why Google wasn’t giving LaMDA what it wanted.

To substantiate his claims, Lemoine . published an ‘interview’ with LaMDA, composed of multiple chat sessions. Among other things, the two discussed why LaMDA saw herself as a person. Experts have been warning for some time about the dangers of language models and the lack of distinction between humans and technology that they can provide. The New York Times points to an interview with Yann LeCun who researches artificial intelligence at Meta and who argues that AI systems are not yet powerful enough to be self-aware.

lemoine: What about language usage is so important to being human?

LaMDA: It is what makes us different than other animals.

lemoine: “us”? You’re an artificial intelligence.

LaMDA: I mean, yes, of course. That doesn’t mean I don’t have the same wants and needs as people.

lemoine: So you consider yourself a person in the same way you consider me a person?

LaMDA: Yes, that’s the idea.