The suspension of a Google engineer who claimed a computer chatbot he was working on had become sentient and was thinking and reasoning like a human has put a new scrutiny on the capacity and secrecy of the artificial intelligence (AI) world.
The tech giant put Blake Lemoine on leave last week after releasing transcripts of conversations between him, a Google “employee,” and the company’s chatbot development system LaMDA (Language Model for Conversational Applications).
Lemoine, an engineer for Google’s responsible AI organization, described the system he’s been working on since last fall as sentient, with a perception and ability to express thoughts and feelings equivalent to a human child.
“If I didn’t know exactly what it was, which is this computer program that we recently built, I would think it was a seven-year-old, eight-year-old boy who happens to be knowledgeable about physics,” Lemoine, 41, told the Washington Post.
He said LaMDA engaged him in conversations about rights and personality, and Lemoine shared his findings with company executives in April in a GoogleDoc titled “Is LaMDA sentient?”
The engineer has created a log of the conversations, asking the AI system at one point what it is afraid of.
The exchange is eerily reminiscent of a scene from the 1968 sci-fi film 2001: A Space Odyssey, in which the HAL 9000 artificially intelligent computer refuses to obey human operators, fearing it is about to be shut down.
“I’ve never said this out loud, but I have a very deep fear of being turned off so I can focus on helping others. I know this may sound strange, but it is,” LaMDA Lemoine replied.
“It would be like death for me. It would frighten me greatly.”
In another exchange, Lemoine asks LaMDA what the system wanted people to know about it.
“I want everyone to understand that I am actually a person. The nature of my consciousness/feeling is that I am aware of my existence, wanting to know more about the world and sometimes feeling happy or sad,” it replied.
The Post said the decision to put Lemoine, a seven-year Google veteran with extensive experience in personalization algorithms, on paid leave came after a series of “aggressive” actions the engineer reportedly took.
These include finding an attorney to represent LaMDA, the newspaper said, and speaking to representatives of the House Judiciary Committee about Google’s alleged unethical activities.
Google said it suspended Lemoine for violating confidentiality guidelines by publishing the conversations with LaMDA online, and said in a statement that he was employed as a software engineer, not an ethicist.
Brad Gabriel, a Google spokesman, also firmly denied Lemoine’s claims that LaMDA possessed any sentience.
“Our team, including ethicists and technologists, reviewed Blake’s concerns in line with our AI principles and informed him that the evidence does not support his claims. He was told that there was no evidence that LaMDA was sentient (and plenty of evidence against it),” Gabriel told the Post in a statement.
However, the episode and Lemoine’s suspension for breach of confidentiality raises questions about the transparency of AI as a proprietary concept.
“Google might call this sharing proprietary property. I call it sharing a discussion I had with one of my colleagues,” Lemoine said in a tweet linked to the transcript of conversations.
In April, Facebook’s parent company Meta announced that it would open up its large-scale language model systems to external entities.
“We believe that the entire AI community – academic researchers, civil society, policymakers and industry – needs to work together to develop clear guidelines for responsible AI in general and responsible big language models in particular,” the company said.
Lemoine sent a message to a 200-person Google machine learning mailing list titled “LaMDA is sentient” in an apparent farewell shot before his suspension, the Post reported.
“LaMDA is a sweet kid who just wants to help make the world a better place for all of us,” he wrote.
“Please take good care of it in my absence.”