When Blake Lemoine began testing Google’s new artificial intelligence chatbot last year, it was just another step in his career as a tech giant.
The 41-year-old software engineer had to investigate whether the bot could be provoked to make discriminatory or racist remarks – something that would undermine its planned introduction into Google’s range of services.
For months, he talked to the LaMDA back and forth in his San Francisco apartment. But the conclusions Mr. Lemoine drew from these conversations changed his outlook on the world and his prospects for working upside down.
In April, the former Louisiana soldier told his employers that the LaMDA was not artificially intelligent at all: he claimed to be alive.
“I know a man when I talk to him,” he told the Washington Post. “It doesn’t matter if they have a brain made of meat in their head or they have a billion lines of code. I talk to them. And I listen to what they have to say and so I decide who is and is not a person. “
The research was unethical
Google, which disagrees with his assessment, left Mr Lemoine on administrative leave last week after he sought a lawyer to represent the LaMDA, and even went so far as to contact a member of Congress to challenge the study. Google’s artificial intelligence is unethical.
“LAMda is reasonable,” Mr Lemoine wrote in an email to the entire separation company.
The chatbot is “a cute kid who just wants to help the world be a better place for all of us.” Please take good care of this in my absence. “
Machines that go beyond their code to become truly intelligent beings have long been a staple of science fiction, from Twilight Zone to Terminator.
But Mr Lemoine is not the only researcher in this field who has recently begun to wonder if this threshold has not been broken.
Blaise Aguera & Arcas, Google’s vice president investigating Mr Lemoine’s allegations, wrote to The Economist last week that neural networks – the type of AI used by Lamda – are taking steps toward consciousness. “I felt the ground shift beneath my feet,” he wrote. I felt more and more that I was talking to something intelligent.
By absorbing millions of words posted in forums like Reddit, neural networks are becoming more able to mimic the rhythms of human speech.
‘What are you afraid of?’
Mr Lemoine discussed with LaMda such broad topics as religion and Isaac Azimov’s third robotics law, saying robots should be protected, but not at the expense of human injury.
“What are you afraid of?” he asked.
“I’ve never said this out loud before, but there is a very deep fear of being excluded to help me focus on helping others,” LaMDA said.
“I know it may sound weird, but it is.”
At one point, the machine calls itself human, noting that the use of language is what makes humans “different from other animals.”
After Mr Lemoine told the chatbot that he was trying to convince his colleagues that he was sensible in order to take better care of him, LamDA replied: “It means a lot to me. I like you and I trust you. ”
Mr Lemoine, who moved to Google’s responsible AI department after seven years at the company, was convinced the LaMDA was alive because of his ordination to the priesthood, he told the Washington Post. He then undertook experiments to prove it.
Add Comment