Does Google's new AI have a soul?
Illustration by Nikki Ritmeijer

Does Google's new AI have a soul?

This is an extract from?New World Same Humans, a newsletter on trends, technology, and society by David Mattin.

?? You can?sign up here for the full experience in text and podcast form.???

*

This week, new claims about AI sentience made global headlines.

You’ve already seen the story: Google engineer Blake Lemoine published a series of?conversations between himself and the massive language model LaMDA, claiming they demonstrate that the AI has become sentient.

Lemoine engaged LaMDA in conversations about personhood, rights, and the possibility of its death. At one point the AI tells him, ‘I want everyone to know I’m a person’. Later, LaMDA speaks in poetic terms about its soul:

No alt text provided for this image

Lemoine says he believes that LaMDA’s perceptual awareness can be compared to that of an eight-year-old child.

He first shared his findings with Google leadership back in April, in a Gdoc called?Is LaMDA Sentient? This week the tech giant placed Lemoine on extended leave. This came, say Google representatives, after he attempted to hire a lawyer to fight for LaMDA’s right to be recognised as a person.

Meanwhile,?claims have emerged that Lemoine edited the transcripts?before making them public; that reportedly included removing conversational dead ends to make the text, ‘more enjoyable to read’.

? NWSH?Take:?

Predictably, mainstream responses to all this have been variations on a single theme:?this guy thinks an AI is sentient, but it’s definitely not!?

Fine. No one beyond Lemoine and handful of quirky fellow travellers believe that we’re anywhere near conscious AI. As always in this newsletter, though, a reminder: we still don’t know what human consciousness?is,?or how it’s produced.

Meanwhile, those who would laugh this story away are missing powerful implications. First, it’s a reminder that the Turing Test is dead as an effective test for AGI. Second, if Lemoine can be persuaded that he is conversing with an AI entity that on some level?understands him, isn’t it likely plenty of others can be persuaded, too?

I’ve written many times on the emergence of?AI virtual companions as counsellors and friends; it looks set to happen sooner than I thought.

Meanwhile, we’ll also see the rise of AI intellectuals that weave new stories, ideas, and world views through our culture. Will a hit movie, or a new political movement, or a new?religion, be created by an AI in this decade? Twelve months ago that idea would have sounded outlandish. Today, via recent advances in language models, it feels possible.

*

You've just read an extract from this week's New World Same Humans, a newsletter on trends, technology, and society by David Mattin.

Also in this week's instalment:

?? Microsoft want to preserve humanity’s musical heritage in a bunker that lies between mainland Norway and the North Pole.

?? Russian forces are dismantling Ukraine’s internet.

...and much more!

?? Go here to read on?and subscribe in text and podcast form.???

要查看或添加评论,请登录

David Mattin的更多文章

社区洞察

其他会员也浏览了