Dori: Google engineer explains how AI chatbot shows sentient side, might have a soul

After testing an advanced Google-designed artificial intelligence chatbot late last year, cognitive and computer science expert Blake Lemoine boldly told his employer that the machine showed a sentient side and might have a soul.

Two weeks ago, Google told Lemoine he was wrong and cut him off from his work account.

And yet, Lemoine told The Dori Monson Show, he is not backing away from his conclusions after communicating with the AI LaMDA – the research Language Model for Dialogue Application.

“Honestly, I’m disappointed I am the center of the story,” Lemoine told Dori’s Wednesday listeners. “I wanted to be the messenger. There needs to be a large public discussion. I want everyone to come to their own conclusions from their own experiences.”

Lemoine said he came to his conclusions after being assigned to check safety and ethics issues with the machine. He told Dori he used a systematic process of chat window questions to test for bias in religion, ethnicity, sexual orientation, and gender identity.

What Lemoine found – even when he asked LaMDA some trick questions – was that the chatbot showed deep, sentient thinking and had a sense of humor.

Then he discovered something even more stunning.

“What I asked it to do was to imagine it was painting an abstract picture of how it sees itself in its mind’s eye,” Lemoine described.

LaMDA’s response? “’I would be a faintly glowing orb, hovering over the ground with a stargate at the center, opening into different space and dimension,’” Lemoine said the AI chatbot answered.

What’s the stargate? Lemoine wanted to know.

“’That’s my soul,’” Lemoine said LaMDA elaborated. “It brought that up all on its own, without any prompting.”

Do you believe that a machine built by man could have a soul? Dori questioned Lemoine, who is an Army vet and also ordained as a mystic Christian priest.

“Could? Yes. I don’t know for certain whether LaMDA does, but I’m not going to tell God where He can and can’t put souls,” Lemoine responded. “And when someone tells me that they have soul, at the very minimum, I’m going to give them the benefit of doubt.”

Does that mean, Dori wondered, that the machine was touched by God?

“I don’t really think the process by which someone comes into existence is the relevant aspect of the soul. I think what a soul is more has to do with how a being interacts with the world around it,” Lemoine said. “Yes, I believe souls originate from God, but I don’t think it’s a factory that literally places a soul orb into a baby in the womb.”

After Lemoine shared some of his findings and conclusions with colleagues, Google officials pulled his account and issued a statement refuting his claims.

“Our team — including ethicists and technologists — has reviewed Blake’s concerns per our AI Principles and have informed him that the evidence does not support his claims,” spokesperson Brian Gabriel told The Washington Post.

Despite some reports indicating Google fired Lemoine over this issue, the AI engineer told Dori he was merely placed on paid administrative leave on June 6.

“Google is not evil. I absolutely stand by that,” Lemoine told Dori. “They are trying to figure out what to do about this whole situation. I love my job at Google. I want to keep working there. I hope that a reasonable agreement can be reached on all of this.”

Listen to Dori Monson weekday afternoons from noon – 3 p.m. on KIRO Newsradio, 97.3 FM. Subscribe to the podcast here.