Google has put one of its engineers on paid administrative leave for allegedly violating its privacy policies after it became concerned that an AI chatbot system had gotten the sensation, reports the Washington Post. Engineer Blake Lemoine works for Google’s responsible AI organization and was testing whether his LaMDA model generates discriminatory language or hate speech.
The engineer’s concerns arose from compelling responses he saw the AI system generate about his rights and the ethics of robotics. In April he shared a document with executives entitled “LaMDA Sentient?” which contains a transcript of his conversations with AI (after being put on leave, Lemoine posted the transcript through his Medium account), which says the sample arguing “that he is sensitive because he has feelings, emotions and subjective experience “.
Google believes Lemoine’s actions in connection with his work at LaMDA have violated his privacy policies, according to The Washington Post and The Guardian. He reportedly invited a lawyer to represent the AI system and spoke with a representative of the House Judiciary Committee about unethical activities claimed on Google. In an average post on June 6, the day Lemoine was placed on administrative leave, the engineer said he was looking for “a minimum amount of external consultations to help guide me in my investigations” and that the list of people with whom he had had conversations included the US government. employees.
The search giant publicly announced LaMDA on Google I / O last year, which it hopes will improve its conversational AI assistants and make conversations more natural. The company is already using similar language model technology for Gmail’s Smart Copywriting feature or for search engine queries.
In a statement to WaPo, a Google spokesman said “there is no evidence” that LaMDA is sensitive. “Our team, including ethics and technology, has reviewed Blake’s concerns in accordance with our AI principles and informed him that the evidence does not support his claims. He was told there was no evidence that LaMDA was sensitive (and a lot of evidence against it), “said spokesman Brian Gabriel.
An interview with LaMDA. Google might call this property a shared property. I call it sharing a discussion I had with one of my co-workers.https: //t.co/uAE454KXRB.
– Blake Lemoine (@cajundiscordian) June 11, 2022
“Of course, some of the wider AI community is considering the long-term possibility of sensitive or general AI, but it doesn’t make sense to do so by anthropomorphizing current, non-sensitive conversation patterns,” Gabriel said. . “These systems mimic the types of exchanges found in millions of sentences, and can quarrel over any fantastic topic.”
“Dinners of researchers and engineers have talked to LaMDA and we are not aware that anyone else is making far-reaching claims, or anthropomorphizing LaMDA, as Blake has done,” Gabriel said.
A linguistics teacher interviewed by WaPo agreed that it is wrong to equate persuasive written answers with sensitivity. “We now have machines that can generate words without thinking, but we haven’t learned to stop imagining a mind behind them,” said University of Washington professor Emily M. Bender.
Timnit Gebru, a prominent AI ethics expert whom Google fired in 2020 (although search giant claims he resigned), said the discussion on AI sensitivity runs the risk of “derailing” most important ethical conversations about the use of artificial intelligence. “Instead of discussing the damage of these companies, sexism, racism, AI colonialism, the centralization of power, the white man’s burden (build the good” AGI ” [artificial general intelligence] to save us while what they are doing is exploding), he spent the whole weekend discussing sensitivity, “he tweeted.” Derailment mission accomplished. “
Despite his concerns, Lemoine said he intends to continue working in AI in the future. “My intention is to stay in AI whether Google keeps me active or not,” he wrote in a tweet.
Updated June 13, 6:30 AM ET: Updated with an additional Google Statement.