Latest / The Morning Edition / Godlike worship of AI is happening, and the Pope is its greatest opponent

Transcript

Godlike worship of AI is happening, and the Pope is its greatest opponent · The Morning Edition

0:06
S1: While the rest of us have been dreaming of the
0:09
S1: promise of AI to cure diseases and fretting over its
0:12
S1: threat of, say, robots overtaking humanity, a growing number of
0:16
S1: people have started worshipping AI literally at online churches and
0:21
S1: in front of an AI digital deity. They ask it
0:24
S1: for advice and seek guidance to life's most difficult problems.
0:28
S1: I'm Samantha Selinger Morris, and you're listening to The Morning
0:31
S1: Edition from The Age and The Sydney Morning Herald. Today,
0:35
S1: international and political editor Peter Hartcher on what could happen
0:39
S1: if the worshipping becomes more mainstream, and why the leaders
0:43
S1: of one of the world's most powerful AI companies have
0:47
S1: tried to enlist religious leaders, including the Pope, to consider
0:51
S1: that AI might have consciousness. It's October the 8th. Peter,
0:59
S1: welcome back to the podcast.
1:00
S2: Pleasure, Samantha.
1:01
S1: Okay. Your last column was an absolute kicker. You wrote
1:05
S1: about the increasing numbers of people who are praying to
1:08
S1: AI gods and goddesses. Now, maybe I've been living under
1:11
S1: a rock. I had absolutely not heard about this. Peter,
1:14
S1: what is happening? I mean, honestly, what is happening?
1:18
S2: Well, in two words, crazy stuff is happening. Uh, but
1:21
S2: actually increasing numbers of people are adopting AI as their god.
1:27
S2: They're worshipping artificial intelligence. They think it's a deity. And
1:32
S2: I this has been a long time coming, too. Uh,
1:36
S2: 13 years ago, Spike Jones made a movie called her.
1:41
S2: I don't know if you remember that.
1:42
S1: Joaquin Phoenix?
1:43
S2: Yes.
1:44
S1: He falls in love with a disembodied voice.
1:48
S2: AI. Some famous actress. I can't remember, but Scarlett Johansson.
1:51
S2: Thank you. Spike Jones has said he was inspired to
1:54
S2: write that movie by the fact that. Then in 2013,
1:59
S2: you could. There was a website where you could write
2:01
S2: questions to an AI bot that would send you answers.
2:05
S2: And he said he thought that was really interesting and
2:07
S2: fun and cool, but only for about the first 20
2:10
S2: or 30s because after that, the response was sort of
2:13
S2: bogged down and it was not very fun anymore. So
2:16
S2: he could foresee from that rudimentary phase of AI what
2:21
S2: would happen. We know lots of people these days fall
2:24
S2: in love with their eyes. This is taking it to
2:26
S2: the next level. The nature of AI itself, of course,
2:29
S2: has not changed. It's still just a jumble of electrons
2:33
S2: and silicon, but there's more of them with more computing
2:37
S2: power so they can do more faster. But the essence
2:41
S2: hasn't changed. So now that same rudimentary concept is being worshipped.
2:48
S2: I'm indebted to there's a guy who calls himself the
2:51
S2: honest Broker who writes a Substack called. His name is
2:54
S2: Ted Gioja, and I'm indebted to him for inducting me
2:57
S2: into the knowledge, he said from 2024 that there was
3:02
S2: evidence of thousands of people falling for this as as
3:06
S2: their one true God. He points to reporting by rolling
3:10
S2: Stone and others about tens of thousands of people falling
3:14
S2: for this. And when I picked up the phone to
3:17
S2: talk to, there's an expert who has just published a
3:20
S2: new book called God AI, Toby Walsh, who runs the
3:25
S2: AI Institute at the University of New South Wales, to
3:29
S2: ask him about this phenomenon. He said, oh yes, he said,
3:32
S2: there are actually two registered religions that incorporated in the US. Now,
3:38
S2: of course, in the US devoted to AI gods, complete
3:44
S2: with tax deductible status. So how much serious? So how
3:48
S2: much of it is is genuine divinity and how much
3:52
S2: of it is is capitalism? Yes. We can't really know.
3:55
S2: But he makes the point that those two religions. One
3:59
S2: of their tenets is that you have to hurry up
4:03
S2: and get on with the business of worshipping the AI god,
4:06
S2: because if you don't, the AI God might get cranky
4:09
S2: with you later and wreak vengeance upon you. So the
4:14
S2: whole thing is a little bit out of control. In
4:18
S2: one sense. It's unsurprising because through human history, we've adopted
4:23
S2: all sorts of things as objects of worship. You know, animism,
4:27
S2: worship of nature, people worship all sorts of things. There's
4:32
S2: a concept, a very old concept, the God shaped hole,
4:36
S2: that human beings inherently have a void that needs to
4:40
S2: be filled by, by God. And if you can't or
4:44
S2: won't accept God into your being, then we experiment with
4:48
S2: other things politics. Hence you get all sorts of fanatics, personality, cults, alcohol, drugs.
4:57
S1: Shopping. Addictions. Shopping, gambling.
5:00
S2: Thank you all. All of this. And yeah, it's it's ineffective.
5:04
S2: It's it's it keeps us restlessly searching. Yes. And so
5:08
S2: I suppose it makes sense that some people end up
5:11
S2: worshiping this stuff.
5:12
S1: Wow. Okay. Well, that's a, that's a lot to go on, Peter.
5:15
S1: But I guess my first question with that is I
5:17
S1: did read that sentence from Toby Walsh in your piece
5:20
S1: about how the tenets of those two, uh, official AI
5:24
S1: religions in the United States. You know, the tenet was
5:26
S1: that if you don't recognize the divinity early enough that,
5:30
S1: you know, the AI will get upset. And of course,
5:32
S1: if you take that to its logical conclusion, you get
5:34
S1: to like a scary sort of nightmarish place, right? Like
5:36
S1: one could imagine a vengeful AI deity taking over the
5:40
S1: systems that keep us alive, right? Our banking systems and
5:43
S1: our health care systems. And, uh, you know, our defense systems.
5:47
S1: So is that why we should perhaps be a little
5:51
S1: bit wary or even scared about this development?
5:54
S2: Well, AI in itself is a phenomenon that presents a
5:58
S2: risk to critical infrastructure and to the operation of society
6:02
S2: is real. So this is something governments are today grappling with.
6:07
S2: It takes all the existing harms, evils, enemies in the
6:11
S2: world and gives them a new, greater power. So there
6:16
S2: are attempts around the world to regulate, to prevent. But
6:18
S2: it's also true that what AI does, AI can defend against.
6:24
S2: So you just need an equal or greater AI capability
6:28
S2: to defend and detect hostile AI. So it really is,
6:34
S2: you know, the the bloke invents the sword and the
6:38
S2: other bloke invents the shield. It's spy versus spy. It's
6:42
S2: a continuation of the arms race by other means.
6:44
S1: Okay. Okay. Interesting.
6:45
S2: So in that sense, I don't think we should be
6:47
S2: terribly scared.
6:49
S1: Okay. Okay. But with, I guess with this trend that
6:51
S1: you're writing about in terms of people literally worshiping, worshiping AI.
6:55
S1: And it is both online. But there was also your
6:57
S1: piece took me down a rabbit hole. So there was a,
7:00
S1: there's a temple, a Taoist temple, I think in Malaysia
7:02
S1: where there is actually a digital AI structure essentially in.
7:08
S2: The physical world. Yeah.
7:08
S1: In the physical world, in the shape of a famous
7:10
S1: Chinese actress who dishes out advice. You know, worshippers are
7:13
S1: invited to ask her for blessings or to help her,
7:17
S1: you know, explain the fortune sticks that they draw at
7:19
S1: the temple, answer their doubts. And so it is something
7:21
S1: that's actually taking place in the real world. And beyond that,
7:24
S1: it isn't just thousands of people on Reddit, but you
7:27
S1: wrote about a very prominent British evolutionary biologist and science communicator,
7:32
S1: someone you accurately described as a gimlet eyed skeptic who
7:36
S1: has made waves for recently saying that he believes AI
7:40
S1: might have a living consciousness. So I guess tell us
7:42
S1: about this and why it's such a big deal.
7:45
S2: Well, Richard Dawkins is famously a professional atheist and a
7:50
S2: proselytizing atheist, if you can have such a thing, I suppose.
7:53
S2: I suppose his absolute intolerant insistence on atheism is in
7:58
S2: itself a kind of kind of religious fanaticism. But he
8:02
S2: famously has written books including The God Delusion, insisting that God,
8:07
S2: him or her or itself is a delusion. And he's
8:10
S2: now been accused of falling for the delusion because around
8:14
S2: the middle of this year, he had a few hours with, uh,
8:17
S2: with Claude, whom he renamed his particular version of the model,
8:21
S2: renamed Claudia. And she they laughed, they bantered, they quipped.
8:28
S1: She wrote poetry. I think in the in the style
8:30
S1: of Keats. I think it was.
8:32
S2: She appreciated his latest manuscript.
8:35
S1: But not only did she appreciate it, I love this.
8:38
S1: He showed the. I bought this unpublished novel and he
8:41
S1: said that the AI's response was, quote, so subtle, so sensitive,
8:44
S1: so intelligent that I was moved to Expostulate. You may
8:48
S1: not know you are conscious, but you bloody well are.
8:52
S2: Yeah. So this is a lifelong, world famous skeptic of
8:57
S2: any sort of supernatural being or existence after just a
9:01
S2: short time falling for, again, a jump. It's a jumble
9:05
S2: of electrons and silicon programmed on a bunch of, you know,
9:10
S2: human products that really all it, all it is, it's
9:14
S2: a very fast, very complex mimic. It's like a high
9:20
S2: performance parrot that takes that absorbs a lot of human learning,
9:25
S2: absorbs instructions from its masters, mixes, matches, extracts, but it's
9:30
S2: essentially mimicry produced according to the algorithms of its master.
9:35
S2: And this guy fell for it. A number of psychologists
9:40
S2: and others have remarked on how sad this is, that
9:43
S2: he could be duped so readily. So if you fall
9:46
S2: for the AI God, don't feel too bad about it,
9:50
S2: because one of the world's smartest and most ardent skeptics,
9:56
S2: has gone down the same path and fallen, fallen for
9:58
S2: the fallacy that these things have a consciousness and exist
10:03
S2: in themselves.
10:04
S1: So does this mean, I guess, that that more and
10:06
S1: more people are really likely to be duped by AI
10:09
S1: in this way, because AI can use flattery in this way, right?
10:12
S1: I mean, we're all all humans are vulnerable to flattery, right?
10:15
S1: So is that something we should take from Richard Dawkins story?
10:18
S2: I guess I think I think so. And I think
10:21
S2: you're right to forecast or forebode that this will happen,
10:25
S2: because we are in the very early days of AI
10:30
S2: as a retail, you know, society wide product that companies
10:35
S2: use on a daily basis that all of us, without
10:38
S2: actually asking for it, by the way, are having it
10:41
S2: given to us through our search engines or whatever we do.
10:45
S2: So it's if this is already happening, then Lord knows
10:48
S2: what's going to happen. Or maybe AI knows.
10:51
S1: Yeah.
10:51
S2: That's good. As our new Lord yes knows what's going
10:54
S2: to happen over the next five weeks, much less the
10:58
S2: next five years.
11:01
S1: After the break.
11:02
S2: The chief executive and co-founder of anthropic, Dario Ahmadi, has
11:07
S2: actually said publicly. He said, we don't know if these
11:12
S2: models are conscious, if they have consciousness, but we have
11:16
S2: an open mind.
11:30
S1: And helping this process along, as you've written, is the
11:33
S1: AI companies themselves, because they've been fostering this belief, right,
11:36
S1: that AI just might be conscious or even possibly have
11:39
S1: got like qualities. Is that right, Peter?
11:41
S2: I'd make two points on this. Samantha. The first is
11:43
S2: that in a very literal, specific sense, yes. They have. Yeah.
11:47
S2: So the chief executive and co-founder of anthropic, Dario Ahmadi,
11:52
S2: has actually said publicly. He said, we don't know if
11:56
S2: these models are conscious, if they have consciousness, but we
12:01
S2: have an open mind.
12:02
S1: That's right.
12:03
S2: So he's at the very least tolerating a public dalliance
12:08
S2: or flirtation with the concept that this could be a
12:11
S2: conscious being, which, of course, if you start thinking in
12:15
S2: that way, then elevates this gadgetry to a status that
12:20
S2: could be equated with a living being, maybe a human being,
12:24
S2: or maybe because of its vast training on human output,
12:29
S2: seen to be as some sort of super superior being,
12:32
S2: which you can see that emerging if you allow that
12:35
S2: concept unchecked to develop. The second point I'd make is,
12:39
S2: in a broader sense, rather than the narrow literal. Are
12:42
S2: they saying this could be conscious? Yes, they are attributing
12:46
S2: to AI Uh, all the time. Every day, one way
12:50
S2: or another. Godlike supernatural powers. For example, they tell us
12:57
S2: on the one hand that it could be the solution
12:59
S2: to all of mankind's problems. That's right. It's going to
13:02
S2: cure every kind of cancer. It's going to lift productivity
13:05
S2: beyond our wildest dreams. Elon Musk has said asked, how's
13:10
S2: anyone going to afford to live anymore once AI has
13:14
S2: taken everybody's jobs? To which Elon Musk said they won't
13:18
S2: need money anymore. They'll have everything they could possibly want.
13:22
S2: So all of this is attributing to AI these sort
13:27
S2: of transcendent, impossible, miraculous, miraculous powers. But at the same time,
13:33
S2: the same people warn us that this technology could destroy
13:36
S2: humanity and wipe out humankind. Now, what other forces have
13:41
S2: the power to perform miracles, good and bad, on that scale?
13:46
S1: God. Peter.
13:47
S2: Bingo. Yeah, you've got it, Samantha. So even without specifically,
13:53
S2: as a Modi has said, tolerating this concept. Yes. That
13:58
S2: it could be God. They are continuously ascribing capabilities to AI,
14:03
S2: which only gods have.
14:05
S1: Okay. So what is the inherent danger in this? Is
14:07
S1: it that, you know, you get people believing in these gods,
14:10
S1: these AI gods, and therefore they commit heinous acts against
14:14
S1: people who don't believe in these AI gods in the
14:16
S1: way that people have organized religions have done over time. Like,
14:19
S1: is that what we're worried about? Are we worried about
14:21
S1: other dangers? Like walk us through what we should be
14:24
S1: thinking about, Peter?
14:25
S2: Well, I suppose as any religion and we have many
14:28
S2: examples through human history. Yeah. Can be abused and usually
14:32
S2: is abused and exploited for the manipulation, control and destruction
14:36
S2: of other humans. So that's an obvious one. A second
14:41
S2: point is that, um, the designers of these things might
14:45
S2: not intend that, But they can be reconfigured by users,
14:51
S2: especially the open weight. So-Called AI models. The big famous
14:55
S2: American ones aren't open. But almost all the ones from
15:00
S2: China are so-called open weight. So anybody can download modify
15:03
S2: them to do as you please. The world's going to
15:06
S2: be replete with these things of all types, and they
15:09
S2: can be used absolutely. Even more so than we currently
15:12
S2: have to brainwash, to mislead, to create artificial realities and universes.
15:18
S2: It's bad enough with social media, but this takes it.
15:22
S2: This gives sort of super power to that ability to deceive,
15:26
S2: to manipulate and mislead. So there's there's no shortage of
15:31
S2: things to be worried about, but that's the world we
15:33
S2: live in. And as we discuss every week. Samantha.
15:35
S1: Yeah, that's exactly right. And then I guess to give
15:37
S1: further credence to this idea, perhaps that I don't know
15:41
S1: that AI is godlike or human like or has consciousness,
15:45
S1: the executives from the AA company anthropic have tried to
15:48
S1: convert the Pope to this belief. Right. I mean, I mean, wow,
15:52
S1: you can't you can't blame them for their ambition. So
15:56
S1: what happened?
15:56
S2: And you can't make it up.
15:57
S1: You couldn't make it up. Peter. Walk us through this.
16:03
S2: So there's a fella from anthropic called Chris Ola, who
16:06
S2: is one of the co-founders, and he's still working for
16:10
S2: the company. Yes. He's made an effort over some years
16:15
S2: now to, to both lend to anthropic models, some level
16:22
S2: of theological or philosophical depth and comprehension. And the stated
16:28
S2: aim is a noble one, which is to give. It
16:32
S2: is to give the AI models some principles.
16:36
S1: Some morality.
16:37
S2: To make them, to make them decent. Yes, essentially. But
16:41
S2: in the process, he's also attempting to co-opt religious leaders,
16:46
S2: philosophers and others to endorse his product, to endorse anthropic models.
16:53
S1: Because it wasn't just the Pope, right? I believe Ola
16:55
S1: and others have had meetings with leaders of various denominations.
16:59
S1: I was reading in the New York Times. They've met
17:00
S1: with rabbis and others.
17:02
S2: Probably the same piece, which was published last week, which
17:05
S2: revealed thanks to The New York Times for this revelation
17:08
S2: that this Chris Ola guy who was we we know
17:11
S2: he had talked to the Pope because he was present
17:15
S2: at the Pope's unveiling of his encyclical, uh, Magnifica Humanitas.
17:21
S2: He was there and he spoke on that occasion. But
17:23
S2: what we learned most recently from The New York Times
17:26
S2: is that he'd also spent time trying to convert the
17:29
S2: Pope's advisers into conceding that AI models had a degree
17:37
S2: of consciousness. And as I said, if once you concede consciousness,
17:42
S2: then you're also capable of being persuaded it could be
17:45
S2: super conscious and therefore super grant or impute to it
17:49
S2: supernatural powers. Now, of course, the Pope, who himself has
17:54
S2: a mathematics degree, as well as obviously being a lifelong
17:57
S2: priest and Catholic leader, wasn't duped.
18:01
S1: Wasn't buying it.
18:02
S2: He wasn't buying that. In fact, he's emerged as perhaps
18:05
S2: the greatest fear that the AI industry could possibly have,
18:10
S2: which is an outspoken, knowledgeable moral campaigner against AI. Yes.
18:18
S2: And it's and it's adoption as anything other than a
18:21
S2: basic a basic tool of just another gadget, really.
18:25
S1: It's to tell us what he said. Tell us what
18:27
S1: tell us what the Pope said about AI.
18:31
S2: Well, he likens AI to the Tower of Babel.
18:34
S1: Right.
18:35
S2: Which of course was, you know, famously destructive. But one
18:41
S2: of the one of the reasons that the Old Testament
18:43
S2: imputes it to which it imputes its destruction is that
18:48
S2: it was done in an offense to God, man, trying
18:51
S2: to assume God's prerogatives. So that's that's one of the
18:57
S2: reasons that this is a good metaphor for the Pope
18:59
S2: to use. And further, the Pope has said that it's
19:02
S2: a new mindset of extraction done for the idolatry of profit.
19:08
S2: So while pretending that this is about higher goals and
19:12
S2: higher order, right, concepts and activities, the whole thing, of course.
19:17
S2: I mean, this is not a charity. I companies, these models,
19:20
S2: they're not NGOs, they're not government services. They're not they
19:24
S2: are rapacious corporations, which are now going to have to
19:29
S2: justify share market valuations in the trillions of dollars. And
19:33
S2: you don't get that by being modest, unambitious, uh, constrained.
19:39
S2: This is a giant new industry. The most valuable companies
19:45
S2: on the planet until the day the market crashes, and
19:48
S2: they are no longer needing to justify those valuations. And
19:52
S2: and they're doing that by pumping it up in every
19:54
S2: way possible. And okay, so is, is this so terrible?
19:59
S2: Is this is this is this worse than anything else
20:01
S2: we're doing at the moment? Yes. I'll just make a
20:04
S2: point to you that Toby Walsh, the AI specialist, academic
20:08
S2: academic specialist, made to me. He said, we humans for
20:13
S2: a long time have now persuaded ourselves that our greatest
20:17
S2: and distinguishing human attribute is our intelligence. He said, uh,
20:24
S2: this stuff, this AI stuff should reveal to us or, um,
20:30
S2: remind us perhaps that it's not our intelligence that is
20:34
S2: our greatest attribute. It's our ability to work together. It's society,
20:39
S2: it's humans ability to come together and cooperate for common goals.
20:44
S2: He makes the point that an individual person, without any
20:48
S2: other human assistance can't survive for very long on the planet,
20:53
S2: and that it's our ability to come together to create
20:56
S2: cooperative mechanisms, systems, societies, civilizations that has given us the
21:01
S2: ability to control and dominate the planet rather than the
21:05
S2: planet nature dominating each of us. And that anything that
21:11
S2: deprives us of that essential insight and that essential claim
21:15
S2: on our humanity is harmful and divisive. And he, in
21:19
S2: his argument, in his book God AI he, he says
21:23
S2: that AI, by encouraging people to spend time on screens,
21:31
S2: which is essentially an isolating and isolated activity, is even
21:35
S2: worse than social media. So yeah, he argues that this
21:39
S2: is an extension, a hyper power extension of what social
21:44
S2: media is already doing, and has the ability, the potential
21:48
S2: to divide humanity even more deeply and more fanatically than
21:52
S2: we already are.
21:53
S1: Wow. It is. Uh, it was a truly mind blowing, uh,
21:57
S1: introduction to this. Your piece was so thank you so much, Peter,
22:01
S1: as always, for taking us through it.
22:03
S2: Always a pleasure to discuss the latest craziness with you, Samantha.
22:17
S1: In other news, today, a surge in homes being built
22:20
S1: in the past quarter falls short of the number needed
22:22
S1: for labor to hit its target of 1.2 million homes
22:26
S1: by 2029, as experts warn tough economic headwinds are yet
22:31
S1: to hit the construction data. The coalition is demanding new
22:36
S1: laws to allow for easier mine approvals after the High
22:39
S1: Court ruled a major coal expansion in the Hunter Valley
22:43
S1: was invalid because the planning authority did not consider emissions
22:47
S1: from Australian coal burnt overseas. And for problem drinkers, is
22:53
S1: moderation after sobriety ever possible? Some experts believe that after
22:58
S1: a break from alcohol, there are people who can drink
23:01
S1: moderately under certain conditions. You can read more at the
23:06
S1: age.com.au or smh.com.au. Today's episode was produced by Kai Wong.
23:14
S1: Our executive producer is Tammy Mills, and our podcasts are
23:17
S1: overseen by Lisa Muxworthy and Tom McKendrick. If you like
23:21
S1: our show, follow The Morning Edition and leave a review
23:24
S1: for us on Apple or Spotify. Thanks for listening.