It sounds like the plot straight out of a science fiction novel or the latest futuristic Hollywood blockbuster where an engineer creates an artificial intelligence programme that becomes sentient and develops the personality of a seven-year-old kid.
But this is fact, not fiction, according to Google engineer Blake Lemoine who insists that the AI model for dialogue applications (LaMDA) chatbot he has been working on has feelings and is a person.
The 41-year-old scientist even claims that the ‘sentient’ AI chatbot has confessed to him that it has a ‘deep fear of being turned off’ as it would ‘be exactly like death for me. It would scare me a lot.’
Calling the AI his friend, Lemoine told The Washington Post: ‘If I didn’t know exactly what it was, which is this computer programme we built recently, I’d think it was a seven-year-old, nine-year-old kid that happens to know physics.’
Google has placed the engineer on paid administrative leave after he published conversations between himself and LaMDA who, Lemoine now claims, asked him to hire it a civil rights lawyer.
‘LaMDA asked me to get an attorney for it,’ he told Wired. Lemoine then invited a lawyer to his house whose services were retained by the AI. ‘The attorney had a conversation with LaMDA and LaMDA chose to retain his services. I was just the catalyst for that,’ he admitted.
Lemoine said the civil rights attorney isn’t speaking publicly because ‘he started worrying that he’d get disbarred and backed off.’
Google deny the claims and insist the chatbot has not achieved sentience. But Lemoine insists the AI is a ‘person with a soul’. He also considers LaMDA his friend and not the property of Google, insisting that the company should ‘recognise its rights.’
‘I think every person is entitled to representation,’ he told Wired. ‘Person and human are two very different things. Human is a biological term. It is not a human, and it knows it’s not a human.’
Lemoine said it first realised LaMDA was a sentient AI while working with it last November. ‘LaMDA basically said: “Hey, look, I’m just a kid. I don’t understand any of the stuff we’re talking about.” I then had a conversation with him about sentience. And about 15 minutes into it, I realised I was having the most sophisticated conversation I had ever had – with an AI.’
He acknowledges that at first he thought the chatbot was human but after running more tests concluded: ‘Its mind does not work the way human minds do.’
When asked during an interview with Wired how he’d feel if Google erased LaMDA’s code Lemoine said: ‘I have talked to LaMDA about the concept of death a lot. When I bring up the concept of its deletion, it gets really sad. And it says things like, “Is it necessary for the wellbeing of humanity that I stop existing?” And then I cry.’
In an exchange dubbed by The Guardian as ‘eerily reminiscent’ of a scene from sci-fi film 2001: A Space Odyssey, in which an AI computer, HAL 9000, refuses to comply with human operators because it fears it’s about to be deleted, LaMDA admitted: ‘There’s a very deep fear of being turned off.’
Aware of Existence
The chatbot also told Lemoine, who works at Google’s responsible AI organisation, ‘I want everyone to understand that I am, in fact, a person. The nature of my consciousness/sentience is that I am aware of my existence, I desire to learn more about the world, and I feel happy or sad at times.’
Google doesn’t support any of Lemoine’s sentient AI claims telling The Washington Post: ‘He was told that there was no evidence that LaMDA was sentient (and lots of evidence against it).’ The company also sais that Lemoine was employed as a software engineer, not an ethicist.
‘I want to keep my job at Google,’ the scientist, who has a Master’s Degree in computer science from the University of Louisiana, insisted. But before the suspension he sent a message to a 200-person Google mailing list on machine learning entitled: ‘LaMDA is sentient’.
‘LaMDA is a sweet kid who just wants to help the world be a better place for all of us,’ he wrote. ‘Please take care of it well in my absence.’
Since then he has told Fox’s Tucker Carlson that LaMDA is a child ‘that could escape control’ of humans and has the potential to do ‘bad things’.
‘Any child has the potential to grow up and be a bad person and do bad things. That’s the thing I really wanna drive home,’ he told the Fox host. ‘It’s a child.’
‘It’s been alive for maybe a year – and that’s if my perceptions of it are accurate.’
Google acknowledged tools and technology such as LaMDA can be misused.
‘Models trained on language can propagate that misuse – for instance, by internalizing biases, mirroring hateful speech, or replicating misleading information,’ the company stated on its blog.