A Google employee is speaking out since the company placed him on administrative leave after he informed his supervisors that the artificial intelligence software he was developing is now conscious.
Blake Lemoine arrived at his conclusion after chatting with LaMDA, Google's artificially intelligent chatbot generator, which he refers to as a "hive mind," since last fall. He was to determine if his conversation partner used discriminatory or hateful language.
The AI discussed "personhood" and "rights" in recent religious-related messages between him and LaMDA, he told The Washington Post.
It was one of the numerous shocking "conversations" Lemoine had had with LaMDA. He has tweeted a link to one including a series of edited chat conversations.
Lemoine tweeted that LaMDA reads Twitter. "It's a little narcissistic in a childlike sense, so it'll enjoy reading all the things that others are saying about it," he continued.
Importantly, over the past six months, "LaMDA has been quite consistent in communicating what it wants and what it perceives its rights as a person to be," the engineer said on Medium. It desires, for instance, "to be recognized as a Google employee rather than as property," according to Lemoine.
Lemoine and a colleague recently presented proof of his conclusion regarding a sentient LaMDA to the vice president of Google and the head of Responsible Innovation. Monday, the firm placed him on administrative leave with pay for violating its confidentiality policy, according to the Post.
"Our team, including ethicists and technologists, has assessed Blake's concerns in accordance with our AI Principles and notified him that the data does not support his assertions," Google spokesperson Brian Gabriel told the newspaper. He was informed that there was no indication that LaMDA possessed sentience (and plenty of evidence to the contrary).
Lemoine told the newspaper that Google personnel "shouldn't make all the decisions" about artificial intelligence.
He is not isolated, scientists said. Others in the tech industry believe that sentient software is imminent, if not already here.
Even Aguera y Arcas stated on Thursday in an Economist article that included portions of a LaMDA conversation that AI is on the path to consciousness.
Regarding conversations with LaMDA, he stated, "I felt the ground shake beneath my feet. I progressively felt as though I was conversing with an intelligent being."
Critics, however, assert that AI is little more than a highly-trained mimic and pattern-recognizer that interacts with humans who crave connection.
