Latest / Uncomfortable Conversations with Josh Szeps / 'Truth' in an Age of A.I.

Transcript

'Truth' in an Age of A.I. · Uncomfortable Conversations with Josh Szeps

• 0:03
  G'day humans, welcome to the safe space for dangerous ideas.
• 0:07
  What will reality feel like when we get most of it via all-knowing machines?
• 0:14
  That's the big doozy of a question as AI comes to engulf so much of our lives.
• 0:19
  Our shared understanding of the world has already been fractured by social media
• 0:24
  and its echo chambers of news and entertainment and misinformation
• 0:28
  and distraction and addiction.
• 0:31
  Could artificial intelligence help re-establish common facts,
• 0:37
  or will it only make the concept of truth even slipperier?
• 0:40
  Those are the questions Steve
• 0:41
  Rosenbaum is grappling with.
• 0:44
  He's an Emmy-winning former television news producer.
• 0:47
  He was a TED
• 0:48
  resident and the boss of the NYC Media Lab.
• 0:51
  He's written a new book entitled The Future of Truth,
• 0:55
  how AI reshapes reality.
• 0:57
  And he joined me to explore the death of traditional news,
• 1:00
  why TikTok is so addictive, how young people think about trustworthiness,
• 1:05
  and how we can fortify facts in a world of slop.
• 1:10
  I hope you enjoy as much as I did the one, the only Steve Rosenbaum.
• 1:21
  Steve, you just got back from TED, from the TED conference in Vancouver.
• 1:25
  And before we were recording, you were saying you had some interesting conversations about truth up there.
• 1:30
  Can you spill the beans on any?
• 1:31
  What's the vibe?
• 1:33
  So at the risk of aging myself, I've been going to TED for a very long time.
• 1:39
  And I actually ended up being a TED resident for five months in their New York offices.
• 1:44
  And so I have kind of a deep seated relationship with the place and the people.
• 1:51
  But every you know, it's there is no room on Earth where there are more incredibly smart people.
• 1:59
  And so, you know, if you know what's good for you, you mostly stay in the corner and don't say anything.
• 2:07
  at the risk of arguing with a PhD in neuroscience,
• 2:11
  which would not have ended well for me.
• 2:15
  The book is not out yet.
• 2:17
  And so I literally had two copies.
• 2:20
  So I couldn't sell them.
• 2:21
  I couldn't give them away.
• 2:22
  I had one in my bag.
• 2:24
  And I trepidatiously went to the bookstore and said,
• 2:29
  can you put this in the bookstore,
• 2:31
  even though it's not for sale?
• 2:33
  And they were generous and said, yes,
• 2:35
  put a QR code on it saying pre-buy.
• 2:37
  So I did that.
• 2:38
  And then I kind of walked around for the
• 2:42
  first day and I kind of said to people, you know, what do you think?
• 2:46
  You know, truth was kind of
• 2:47
  already floating in the universe.
• 2:51
  Ted had asked me to host a truth dinner and the Ted dinners are
• 2:58
  often very kind of free willing.
• 3:01
  So I showed up at this restaurant and there was a massive sign
• 3:06
  that said, you know, the future of truth on this restaurant.
• 3:09
  And I looked at and there were 80
• 3:11
  tables, 80 seats.
• 3:14
  And I was like, holy shit, like nobody told me this was a thing.
• 3:19
  And it turned out
• 3:19
  it was sold out.
• 3:21
  And and just to be clear so that your listeners don't tune out, not because of me,
• 3:26
  Not, oh, I want to go see Steve Rosenbaum talk, but because of the sub, the word, not even the subject, but the word.
• 3:34
  And the dinner was.
• 3:35
  You're talking about the word the.
• 3:37
  Yes, the word the.
• 3:38
  Oh, the word truth.
• 3:39
  Got it.
• 3:39
  The word truth.
• 3:41
  The dinner was fantastic.
• 3:45
  The conversations were unbelievably riveting.
• 3:47
  I walked around and listened to the various tables.
• 3:49
  And then at the end, each table got up and talked about for a moment what they engaged in.
• 3:55
  It was so wide ranging and fascinating.
• 4:00
  And then the next day, something happened to me that has never happened in my entire professional life, which is I ended up back on the floor in the circle at TED.
• 4:10
  And I was, for a moment, a rock star.
• 4:16
  And people would come up to me and be like, we loved your dinner.
• 4:19
  It was so interesting.
• 4:20
  And I'm just like, you know that had nothing to do.
• 4:24
  I didn't even do it.
• 4:25
  I injected the word truth into a room.
• 4:28
  But this is this moment in time that we're in and why I'm delighted to be talking to you.
• 4:33
  Because there are people who believe that there are a bunch of different positions that we can talk about.
• 4:42
  One is we're in a post-truth world.
• 4:44
  Truth is over.
• 4:45
  You know, statistically by.
• 4:47
  And I examine that in the book.
• 4:50
  I don't want to give away the book's answer.
• 4:52
  But post-truth, I have an interesting dialogue with the keeper of the post-truth flame.
• 4:58
  And I think I come out slightly ahead, ever so slightly, which is no, not accepting post-truth as an outcome.
• 5:09
  But then there's another argument, which is all this tech is basically going to eat humanity and there's really nothing to do about it.
• 5:16
  and we should just go about our business and wait for Andrew Yang and UBI to come and pay our bills.
• 5:27
  And I don't think I'll end up there either.
• 5:32
  But, Josh, I'll put you on the spot.
• 5:36
  I mean, you do this every day.
• 5:39
  when you think about your work is there a better way to describe you than a truth teller
• 5:48
  there would be less pretentious ways uh i wouldn't i wouldn't choose the moniker
• 5:52
  myself uh but i think you're pointing at something uh important which is that i mean there's a
• 6:00
  distinction that i'd be interested in because i i didn't get it clearly necessarily from the book
• 6:07
  or from your other conversations that I've heard between facts.
• 6:14
  And so there are facts and then there are opinions,
• 6:17
  but then there's a squishy middle that I don't think gets necessarily
• 6:21
  enough attention and that I'm not quite clear on how you wrestle with.
• 6:26
  Like my relationship with the public broadcaster fell apart in part
• 6:31
  because I thought that there were areas where everyone sort of agreed on the data,
• 6:40
  but there were sufficient differences in how people might interpret what is right to glean
• 6:46
  from the data and what it means to be on the right side of history that would veer some people
• 6:53
  towards emphasizing different interpretations over others.
• 6:57
  So if you think about COVID,
• 6:59
  um you know there are facts about vaccine injuries there are facts about lockdowns there are facts
• 7:06
  about the economic impact that all put me firmly on the side of being uh you know we did the best
• 7:13
  that we could and vaccines are safe and they work and so it was good to try to get as many people
• 7:19
  vaccinated as we could um but a person can look at those exact same facts and still say in their
• 7:26
  in their opinion, the human rights violation of cajoling so many people into getting vaccinated
• 7:34
  when they would probably have been okay if they caught COVID was an error and that the lockdowns
• 7:41
  were overreach.
• 7:42
  You can go through this on a million different things, on transgender issues
• 7:45
  or any of the other culture war stuff where, yes, there might be some disagreement around the edges
• 7:50
  about fact claims that are actually legitimately verifiable using data.
• 7:56
  How many people under the
• 7:58
  age of 18 are getting puberty blockers?
• 8:00
  How many people are getting physical surgery at what ages?
• 8:04
  How many people regret it after the fact?
• 8:07
  But then there are kind of values questions about
• 8:11
  whether or not we sort of believe that it's desirable for people's gender to be fluid,
• 8:16
  or whether we believe that it's desirable for there to be traditional gender roles that will
• 8:20
  skew people's skew the way that people report on this and that for me is where the concept of truth
• 8:28
  becomes more malleable am i chasing a red herring there no but i want to even before we get i'm happy
• 8:36
  to carve up your red herring but before we do that let's go back a step so it used to be all
• 8:43
  kinds of things were presented to us as kind of fait accompli.
• 8:47
  And if we thought deeply about
• 8:52
  any of them, you could make a civil rights argument that motorcycle helmets were an invasion
• 8:58
  of my privacy and not being able to smoke in bars was an invasion of my privacy.
• 9:02
  And, you know,
• 9:03
  I have a personal thing about traffic lights.
• 9:06
  I think they're an invasion of my privacy.
• 9:08
  I want
• 9:08
  to be able to drive.
• 9:09
  If it says red or green, fuck the color.
• 9:11
  I want just people to go through it.
• 9:12
  The only thing is one out of every 10 times you do that, you die or you kill someone.
• 9:19
  And so as a society, we make decisions about COVID and vaccines.
• 9:25
  And we used to make those decisions kind of at a somewhat in a cocoon.
• 9:31
  And then the government or traffic lights or helmets would present them.
• 9:36
  What changed was two things, social media and a business model that said it is really good for these platform companies to literally rile people up, to just to take the most extreme, disagreeing, angry points of view and amplify the shit out of them.
• 9:56
  so that it seems like everybody's debating whether or not vaccines are good for you.
• 10:00
  When in fact, if you dig into the data there, what you find is it's four people,
• 10:05
  many of whom aren't in either of our countries, but are sitting in a boiler room in Russia,
• 10:11
  doing it just because being a provocateur is profitable as hell.
• 10:19
  And that leaves average people, I think, pretty exhausted.
• 10:26
  when in fact, like nobody wants to take a vaccine.
• 10:29
  Nobody wants a shot.
• 10:31
  But at the end of the day, I took it.
• 10:34
  I'm glad I did.
• 10:35
  I didn't get COVID.
• 10:36
  I don't believe it was herd immunity that saved me.
• 10:40
  Wait, you didn't get COVID?
• 10:41
  I did not get COVID.
• 10:43
  You're the one.
• 10:44
  I am the one.
• 10:45
  But what's important is like the number of messages
• 10:50
  that were coming at me as a New Yorker
• 10:53
  telling me that it's this crime against freedom and i'm you know damn the man and i should not be
• 10:59
  forced to you know and and we can you and i now consider and have this conversation philosophically
• 11:08
  because we're both not dead harder to have this conversation if we were dead well that's true i
• 11:15
  mean of course but i mean it's on to be fair to the other side it's unlikely that either of us
• 11:19
  would have died getting COVID unvaccinated if we weren't in the demographics that were
• 11:24
  most at risk.
• 11:25
  I mean, I mainly got it because I just wanted everyone to have it so that people
• 11:30
  who were much more vulnerable than me would have slightly less community spread and we could all
• 11:34
  go back to living normal lives.
• 11:37
  Not so much because- By the way, I love that you say that,
• 11:40
  like that was a fait accompli.
• 11:42
  That already makes you a more optimistic person than I am.
• 11:45
  I thought
• 11:45
  we were going to be trapped in our apartments living on Zoom for the rest of our lives.
• 11:48
  Well, we had different experiences.
• 11:50
  I mean, New York, you know, went through a whole trauma in early 2020 that we were able to escape.
• 11:57
  We bagged parts out the window.
• 11:59
  Absolutely, yes.
• 12:01
  And so, yeah, sorry, carry on.
• 12:03
  But I mean, are you saying that it was a good thing?
• 12:06
  So obviously there was this other era when everyone tuned into Walter Cronkite and, you know, some middle-aged white straight male talked at people for half an hour every evening.
• 12:15
  and that was all of the information that you got
• 12:17
  and people were perhaps more unquestioning.
• 12:20
  I mean, the splintering of...
• 12:23
  And all three networks broadcast the exact story rundowns.
• 12:27
  They looked at each other during the day and they said,
• 12:29
  oh, I know this because I was an intern at CBS back in the day.
• 12:34
  And they would look at each other's network news broadcast
• 12:37
  and called each other and then the shows were identical.
• 12:41
  I mean, it wasn't like you could tune in.
• 12:44
  there was no fox news there was no you know um not that i necessarily feel like i watched those
• 12:53
  would have made news like there was some there's something soothing about a moment where you wake
• 13:00
  up in the morning you put on your clothes you brush your teeth you take your kids to school
• 13:04
  you go to work you come home you watch the news for half an hour you read the paper but that's it
• 13:09
  Like whatever is going on in the world isn't coming at you on your phone 24 hours a day, seven days a week.
• 13:16
  And now.
• 13:16
  Right.
• 13:17
  But I mean, so the champions of the demise of that status quo and of the new information ecosystem, be they people on the left like Glenn Greenwald, be they activists like Black Lives Matter or, you know, the trans community, or be they people on the right like Tucker Carlson, would say that that excluded a whole bunch of different voices and allowed rich and powerful institutions and, you know, organizations, the deep state and the white people.
• 13:46
  patriarchy or whatever to impose a worldview on everybody else and it excluded marginalized
• 13:52
  voices and it allowed calamities to be carried out in our name without proper oversight
• 13:58
  um and that now we have a much more disaggregated system where i mean this was i think part of what
• 14:04
  led to what you might be called what might be called wokeness or or cancel culture uh you know
• 14:09
  between 2012 and 2022, perhaps, where all of a sudden the reins of power over what was acceptable
• 14:17
  from the mainstream media were wrestled back into the hands of people who anointed themselves as
• 14:23
  the people and imposed new standards and new norms that were, in hindsight, probably too aggressive.
• 14:30
  But at the time, felt like they were a way of spreading truths that had previously been
• 14:37
  unasserted or sidelined?
• 14:39
  So we're midway through a journey.
• 14:44
  And so you're choosing to stop us at
• 14:46
  a moment, which is fine.
• 14:48
  And that moment is what was labeled, I think, brilliantly and early on
• 14:55
  as the filter ball.
• 15:00
  We've now used enough words in this conversation already that listeners who
• 15:06
  don't agree with either you or i politically have already had they stumbled into us by accident
• 15:11
  they're gone right i mean probably the moment you said trans they were gone
• 15:18
  uh and maybe i mean it depends it depends whether they are coding my reference to that as being
• 15:25
  partial to one side or the other but i i don't know each other that well but i'm going to say
• 15:30
  you are i i mean well probably probably my saint my belief my claim that everyone should have
• 15:36
  gotten vaccinated is probably the the the most partisan thing that i've said that is but it all
• 15:40
  lines up and whatever but my point is and this by the way i think is is why we need to wave a flag
• 15:48
  about where we are now and where it's going because if the amount of information that's coming
• 15:54
  at people is overwhelming and unpleasant and not providing either joy or knowledge,
• 16:03
  then really tuning out is a really good answer to that.
• 16:08
  Like, you know what?
• 16:10
  I don't need to
• 16:10
  listen or watch any of this stuff.
• 16:12
  I'm just going to watch, you know, copy bars on TikTok.
• 16:16
  they're both lovely uh and and the algorithm that larry ellison now controls unclear how
• 16:25
  much his hands are on the dials but i would guess as much as they you know as like meat hooks you
• 16:31
  know i mean there will be more copy bars per square inch or whatever kind of silliness um
• 16:39
  kind of addictive silliness um but but tuning out news and information i think could be argued as a
• 16:48
  an act of self-preservation at this point and does that relate and so the i mean that's one
• 16:56
  response that people have another response is that they go to the most extreme versions as you
• 17:02
  alluded to earlier of news and information and obviously the algorithms are disastrous in the
• 17:07
  fact that they will serve us things that either reinforce what we already believe or they'll
• 17:12
  demonize what we don't believe but they'll keep us in the echo chamber one way or another because
• 17:17
  we're likely to keep scrolling and commenting and liking and sharing if they do that um does that
• 17:24
  relate how does how does that relate specifically to our grounding on truth if we're not extricating
• 17:31
  ourselves from what's going on but we're sort of inhabiting silos so let's go back to the where you
• 17:38
  began this piece of the journey because i think it's relevant um uh objectively truth as a as a
• 17:48
  word in in our trying to figure out the planet is a is a shitty word and it's a shitty word because
• 17:55
  as you pointed out wisely it really means two entirely different things i mean there's objective
• 18:01
  truth and subjective truth and they are it's like saying going to a restaurant and saying
• 18:05
  i want either ice cream or steak but we want to call them the same i mean you know scientific
• 18:12
  truth which is based in facts one plus one equals two the sun rises in the west and sets in the east
• 18:17
  all that stuff is is proved i hate to break it to you but the sun rises in the east and
• 18:22
  it sets in the west but by the way i by a little scientific dyslexia sorry uh
• 18:30
  And then there's, you know, Josh is a great podcaster, which for some number of your listeners is absolutely true.
• 18:38
  Also factually true.
• 18:39
  Yes, that's right.
• 18:41
  And for smothers, they're out of here.
• 18:43
  Like, you know, they've given you 15 seconds and they're off in search of Tucker Carlson.
• 18:50
  And until fairly recently, you could have a barroom conversation about truth and not get caught in the in the vice of that problem.
• 19:00
  And I'll just I'll give you an example.
• 19:03
  So I'm in Vancouver and there's no phones in the auditorium.
• 19:08
  There's no.
• 19:08
  So, like, I'm literally not online for five days, but I'm also kind of a little bit of an addict.
• 19:14
  So I sneak away to check the news and or whatever I'm checking.
• 19:18
  I guess it was TikTok.
• 19:20
  And there's this the first video that comes up is a video of the Artemis space mission.
• 19:27
  And they're showing the three astronauts hanging in front of wires in front of a green screen.
• 19:34
  And they're they're suggesting that the whole thing was staged.
• 19:39
  And I think, yes, very silly.
• 19:41
  And then there's another Artemis right after that is another Artemis video.
• 19:45
  And they're showing that her hair wasn't floating in the air the right way.
• 19:49
  And I'm like, OK, so there's two in a row suggesting that the Artemis mission is staged.
• 19:56
  And then the third one is Jimmy Kimmel on set talking with great passion about the staging of the Artemis mission.
• 20:07
  And it becomes clear to me that I don't know if any of those videos are real or whether that is really something Jimmy Kimmel put on television or whether somebody manufactured a fake Jimmy Kimmel in front of a fake evening, you know, late night show.
• 20:23
  And I literally threw my hands in the air and was like, because I was pretty sure we went around the moon.
• 20:29
  But a little less after that.
• 20:32
  and i mean the the relevant question here is steve why are you sneaking away from the ted
• 20:37
  conference to go on tiktok yeah for writing a book about truth because i'm a little addicted
• 20:44
  oh man that shows us how how far we've fallen uh i assume the kimmel i assume the kimmel clip was
• 20:50
  a high unless he was doing a bit the fact that i had to go through that process i mean i'm a
• 20:59
  reasonably well-read, reasonably
• 21:01
  visually literate person.
• 21:02
  I've made
• 21:03
  television for a chunk of my career.
• 21:05
  And the fact that I
• 21:07
  came away from that
• 21:09
  literally rattled.
• 21:12
  I mean, not
• 21:14
  rattled to my core,
• 21:16
  but
• 21:16
  I wrote a
• 21:19
  piece maybe
• 21:21
  four weeks ago about
• 21:23
  Sora 2,
• 21:25
  in which I argued that it was
• 21:28
  singularly the most
• 21:29
  dangerous piece of technology I'd ever seen.
• 21:32
  And just explain what Sora 2 is.
• 21:34
  Well, what it was, it's been killed.
• 21:37
  But we'll arrive again.
• 21:40
  It's a video app that
• 21:43
  ChatGPT, OpenAI spun out as a social network that allowed you to create totally believable
• 21:50
  videos from a whole clock.
• 21:51
  So I can very easily say, I'm going to take Josh's head, I'm going to
• 21:55
  put it on a body.
• 21:57
  I'm going to have Josh say something that he
• 21:59
  would never say, and I will distribute it in this so-called social network, but it will
• 22:03
  be pixel perfect.
• 22:08
  I interviewed Chad GPT about it
• 22:12
  for about two hours.
• 22:14
  What we came to at the end
• 22:15
  of two hours... You interviewed a human at OpenAI, or you interviewed Chad GPT?
• 22:20
  I interviewed the LMI.
• 22:23
  The bot, right.
• 22:24
  Yeah.
• 22:24
  And what we came down to was that it was dangerous,
• 22:29
  that it was a bad business model.
• 22:32
  Sora, this is.
• 22:33
  Sora.
• 22:34
  That it had damaged OpenAI
• 22:37
  and that Sam should have known better.
• 22:41
  Right.
• 22:41
  But to be fair to an LLM,
• 22:44
  it's going to tell you what it thinks you want to know, right?
• 22:47
  I mean, if you ask it the right prompts,
• 22:49
  you might have been able to get the opposite conclusion.
• 22:52
  They shut it down a week later, so I felt like I was...
• 22:55
  You have indicated.
• 22:58
  But we're still on the plane of data, of what actually happened.
• 23:02
  Like the Artemis astronauts were either up there or they weren't.
• 23:05
  They were either in front of a green screen dangling on wires or they weren't.
• 23:09
  The question of how much attention we should pay to the Artemis mission,
• 23:15
  whether or not, you know, how we should understand it,
• 23:17
  how we should interpret it in the context of the American space race with China
• 23:22
  in the context of whether resources should be spent
• 23:24
  on space exploration at all.
• 23:27
  All of those things used to be things that would come
• 23:30
  through a filter that served the powerful,
• 23:35
  or at least the people who were making the news,
• 23:37
  and now those conversations are democratised,
• 23:41
  and that is confusing and chaotic, and perhaps people will go
• 23:49
  into their echo chambers of interpretation,
• 23:50
  even if they do agree about the facts.
• 23:55
  Does that still fall under the banner of different truths?
• 24:00
  I have to parse that for me.
• 24:02
  There are like four questions, though.
• 24:04
  No, no, it's just one question.
• 24:06
  Are you limiting the definition of truth
• 24:10
  to things that actually happened,
• 24:13
  or does it include what we all agree is true
• 24:17
  about our interpretations of what happened?
• 24:19
  Let me punt that question and instead tell you a story that will get toward the answer of the story.
• 24:26
  So here's how the book got written, because it's a weird journey.
• 24:31
  So I tend to work 80, 90 hours a week.
• 24:37
  I am back when the book began running the media lab for NYU with six universities.
• 24:44
  And I'm enjoying my job.
• 24:45
  It's busy.
• 24:46
  But I'm working 35 hours a week.
• 24:48
  And my wife, as COVID arrives and we end up locked in our apartment, says to me, you're going to go insane.
• 24:57
  You should go get a master's degree to have something to do with your idle brain.
• 25:03
  And so I, because I was at NYU, stumbled into this program called Gallatin, which is a self-determined major.
• 25:10
  And as I'm figuring out how to apply, my then application coach says to me, what interests you?
• 25:17
  And I say, having spent my whole life in journalism or storytelling or filmmaking or magic, which is another thing, truth.
• 25:26
  And so I write a thesis proposal to study truth.
• 25:32
  Midway through odd coincidence, COVID and my master's arrives the very first wildly public LLM.
• 25:44
  And all of a sudden, it's clear to me that this thing that I'm studying, I'm literally reporting on a chapter in which it's changing before my eyes.
• 25:57
  And so the book is, among other things, an interview with eight brilliant people, far smarter than I am.
• 26:06
  Just to be clear for your listeners, didn't get a PhD in truth, got a master's in truth.
• 26:12
  So I am not Dr.
• 26:14
  Truth.
• 26:15
  I am.
• 26:16
  And despite my mother's wishing I was Dr.
• 26:18
  Truth.
• 26:20
  But but part of what I think makes the book fun and our conversation fun as we get into it is like.
• 26:29
  All of these AI is going to come and fill in the blank, this thing, art, music, theater.
• 26:36
  Politics, drama, news, information, all has interesting stories around it.
• 26:42
  If you wrote a book about any of those things, no one would read it.
• 26:49
  Because, you know, AI and law would be read by a microcosm of lawyers who care.
• 26:55
  You know, AI.
• 26:56
  But when you make it a journey, which is what I am pretty good at, each of these characters.
• 27:02
  So, for example, there's a chapter on how AI is impacting warfare.
• 27:09
  Could have written a book on that.
• 27:10
  Not could have written a book on that.
• 27:12
  But when you think about the debate about whether or not there should be a human in the loop, which is kind of the pushback, like it's so patently absurd.
• 27:24
  Given the speed of decisions automobiles make or bombs make that we aren't going to hand those decisions over to a robot.
• 27:34
  Like, when you get to the end of the book, what I think I'm able to argue is that where AI is going to take us as a species, this is not just television after radio or the printing press.
• 27:53
  This is not that kind of change.
• 27:56
  This is something really deeply more fundamental.
• 27:59
  And that makes it worth your time and my time to think about.
• 28:04
  In what way is it more deeply fundamental?
• 28:11
  Because humans have a boiling point where, you know, because we're biological creatures, we need, for example, sleep.
• 28:24
  and there's no pill that makes sleep go away other than, I guess, coffee, lots of coffee for a while.
• 28:31
  So at the point at which the volume of information coming at us exceeds our ability to manage it,
• 28:39
  we become willing victims of AI coming in and saying, listen, Josh, you know, you've got this
• 28:49
  cold and you've asked chat GPT how to get this medicine and chat GPT said, you know what?
• 28:56
  We
• 28:56
  know where there's a perfect cold medicine for you.
• 28:59
  We know the name of the pharmacy.
• 29:01
  We've got
• 29:01
  a deal with the delivery service.
• 29:03
  We can take care of the whole, we can have it at your doorstep in
• 29:06
  20 minutes and you feel better, but you didn't make a decision.
• 29:13
  You basically said to the robot,
• 29:15
  give me a medicine and the robot didn't say to you, by the way, you're not paying me.
• 29:21
  So you're not the customer.
• 29:23
  You're the product.
• 29:24
  And the drugstore is paying me and the medicine
• 29:27
  company is paying me.
• 29:29
  So really, you're not getting the best treatment.
• 29:31
  You're just getting
• 29:31
  the one that's paying the most for your illness.
• 29:35
  And that's kind of happening right now.
• 29:38
  And I mean, that's not happening yet.
• 29:41
  It's not happening on the LLMs that I use.
• 29:42
  That experience
• 29:43
  hasn't happened to me where...
• 29:45
  Have you tried to buy a plane ticket?
• 29:48
  They haven't been very good,
• 29:49
  so I haven't pursued it.
• 29:54
  Why?
• 29:54
  Are the airlines doing deals with the LLMs?
• 29:59
  Well,
• 30:02
  everybody in any form of retail
• 30:05
  is doing...
• 30:07
  I mean, I have a friend
• 30:11
  who's at an investment bank
• 30:13
  and he told me with great excitement this morning
• 30:16
  that if you go on to one of the LLMs
• 30:18
  and ask the best investment bank in this category in New York,
• 30:21
  it recommends them.
• 30:24
  And I said, how did that happen?
• 30:27
  And he said, we have no idea.
• 30:30
  I mean, it's replacing Google and search
• 30:35
  in a way that is in some ways kind of invisible.
• 30:40
  and to your point the right question you're asking is is it yet an economic transaction
• 30:45
  the answer is not quite yet but when you look at how much money sam has borrowed
• 30:49
  and how much money he owes investors like there are only so many places you get money
• 30:55
  sam open ai i mean sorry i just don't call him sam because that's right first like yeah i mean
• 31:01
  hang on i think there are a few things going on so the amount of money that these companies are
• 31:06
  valued at in order for them in order for their shareholders to ever make a profit on that in
• 31:14
  terms of actual returns rather than treating the share as a as a gambling tool i'm assuming it'll
• 31:18
  go up but in terms of actual revenue coming in is what you're saying is that could only take place
• 31:26
  if they did massive deals with advertisers and started pushing you products in their llms the
• 31:31
  alternative explanation is it's a bubble um in the stock market and uh those prices are going to come
• 31:36
  down there's no way that they're going to meet those revenues and whatever revenue they could
• 31:41
  get from advertisers they could also get from subscriptions and corporate subscriptions from
• 31:47
  people who want an unsullied llm that isn't constantly shoving them uh you know a particular
• 31:52
  product at a particular pharmacy that's a few blocks away as we're seeing with the decline of
• 31:56
  of conventional search i mean a lot of people you know you google is not what google used to be
• 32:01
  it used to be because even set aside the ai gemini uh you know uh summary that you get at the top of
• 32:08
  your google search just thinking about that um did you did your computer just bing when i said
• 32:13
  gemini by the way gemini oh i heard something either uh and you know the the amount of sponsored
• 32:21
  searches that you have to scroll through uh even on amazon to just get to a legitimate product is
• 32:26
  and this is a known quantity of something that we have to deal with in the slop of the internet
• 32:31
  these days so there's the the overvaluation of the companies there's the looming question of
• 32:37
  whether or not they will start undermining their llm conversations by inserting uh paid advertising
• 32:45
  into it but well just park those for a moment hold on before we pass that though like yeah we've all
• 32:51
  used it well enough to know that it's already you know gary marcus who's in the book talks about
• 32:58
  them as stoic parrots i mean they already deliver bizarrely inaccurate information seemingly at when
• 33:05
  they when they feel like it like like i could literally ask it to give me 10 quotes from you
• 33:12
  and five of them would be shockingly accurate you'd be like how did it know i said these things
• 33:17
  and then two or three would be totally made up they'd sound like they'd almost be you but we've
• 33:22
  all had that happen where we've asked them to play back some things we've said in public and then be
• 33:27
  Like, I didn't say that.
• 33:29
  Yes, there's definitely an accuracy problem.
• 33:31
  There's definitely a hallucination problem.
• 33:34
  A lot of AI experts think that those will just get...
• 33:36
  By the way, people who are less kind
• 33:36
  might replace the word hallucination with lying.
• 33:39
  But just saying.
• 33:42
  Well, lying implies that it has an agenda.
• 33:44
  I'm not sure it has an agenda in making up my quotes.
• 33:48
  It hasn't.
• 33:49
  By the way, I'm going to circle that one.
• 33:51
  Does it have an agenda yet?
• 33:54
  Isaac Asimov should wander into the frame.
• 33:57
  exactly that's right uh you have no understanding of what uh of what agendas computers could have
• 34:04
  um the i mean the so the yes it definitely i mean i've had weird hallucinations even recently
• 34:10
  they've certainly gotten better but they actually the main hallucination the most egregious
• 34:14
  hallucination that i've had lately which is interesting steve was is that i was talking
• 34:20
  i was preparing you know sometimes if i'm inadequately prepared for an interview that
• 34:24
  I'll just turn on ChatGPT's voice chat function when I'm in the car on the way to the studio.
• 34:30
  I'll just pick up some other, throw around some ideas about what I might talk about.
• 34:34
  And I was interviewing a former political figure about something to do with American politics.
• 34:40
  And I was talking to ChatGPT about it and said, was this person involved in the second Trump administration?
• 34:51
  And ChatGPT started talking about how Donald Trump didn't win the election in 2020.
• 34:58
  And I said, I'm talking about the current Trump administration, not Trump's first term.
• 35:04
  And ChatGPT was emphatic that Donald Trump did not have a second term.
• 35:11
  And I said, that's not true.
• 35:13
  Donald Trump is the president of the United States right now.
• 35:15
  And ChatGPT said, that's incorrect.
• 35:17
  the current president of the united states is joseph robin of biden um and this was you know
• 35:25
  a month or two ago yeah and uh i've had it i've had that happen too and you know when you when
• 35:33
  you're and if you get caught up in a conversation with it you can think well if i correct it i'm
• 35:39
  somehow fixing the training data and it's at least learning that that's a mistake but it's not
• 35:45
  it's you're not training there so you could tell it no he's really president right now i swear to
• 35:49
  god and it says no and you just get stuck in this loop where you're like yeah then you have to turn
• 35:53
  it off and skip over to claude or something where you know you just have to have to close the the
• 35:59
  circuit breaker but i when i reflected on that steve i wondered what was happening and why that
• 36:03
  was so much worse than any other argument that i'd ever had with chat gpt and so much more factually
• 36:08
  wrong about a fact than anything else that chat gpt has said in at least eight to ten months
• 36:15
  where because it's getting a lot better and my suspicion and you might disagree with this my
• 36:19
  suspicion is that in order to counter uh trump's election claims that the 2020 election was rigged
• 36:29
  open ai has put its finger thumb on the scale on chat gpt and said make sure you understand
• 36:38
  that Trump lost the 2020 election
• 36:40
  and that you don't give credence
• 36:42
  to people who say otherwise.
• 36:44
  And that ChatGPT has gone a bit bonkers on that
• 36:47
  and has misinterpreted that
• 36:49
  and got confused when I was talking to it
• 36:51
  because it seemed much worse than anything else.
• 36:53
  So that may not be a natural product of the LLM,
• 36:56
  but maybe a product of people trying to insert truth
• 37:00
  into the LLM.
• 37:02
  So important two things I want to say about that.
• 37:06
  First of all, I am not a developer
• 37:07
  and I'm certainly not a large language model expert,
• 37:11
  but I have friends that are.
• 37:13
  And what they tell me,
• 37:15
  as long as I don't use their names in public,
• 37:18
  is that some of the things it does
• 37:21
  are completely baffling to them.
• 37:25
  Like as a non-computer person,
• 37:27
  there's a temptation to think
• 37:28
  that when something's broken,
• 37:31
  if you're taking a road, a trip,
• 37:32
  and it's taking you down the wrong road,
• 37:34
  you can just go in and write a line of code that says,
• 37:35
  no no not a right turn at that corner you can fix it but lms don't work that way it's relational and
• 37:42
  and very almost slightly mystically strange that's so so how so if you say to them explain to me why
• 37:50
  this mistake is being made i don't think it's as simple as a thumb on the scale i think they wish
• 37:55
  they could put a thumb on the scale no i i take that i take your point that it is weird that
• 38:00
  they're growing things you know that are sort of you know these creatures are kind of clambering
• 38:04
  out of a pandora's box and they don't quite know how but it is also true that they can put guard
• 38:08
  rails around it i mean just try to ask it you know how to commit suicide or try to ask it um
• 38:14
  you know if it can explain to you i don't know why uh you know uh i don't know islam is an evil
• 38:22
  religion or something else that you know and it'll it'll say well we're not gonna we have to be
• 38:27
  careful about that and i'm not allowed to do that or i'm you know so there are there are ways in
• 38:31
  which it knows that it has to stay within certain parameters so my experience is it'll say if i ask
• 38:38
  it a question it doesn't want to answer it'll say no and if i come back forcefully and say you can
• 38:45
  answer that i'm a journalist or you can answer that i'm ready it will then go and answer it
• 38:50
  like oh really no that's not my i i don't know how to get around it i was writing i was writing
• 38:55
  a script and i wanted to come up with some i asked it for a name i was like there's a very
• 39:00
  flamboyant uh sort of gender non-binary uh character uh who is you know sort of traditionally
• 39:08
  camp and you know prone to exaggeration and uh and flamboyance and can you come up with a list
• 39:14
  of 10 names for it and it wouldn't do it it was like it's important that we don't caricature
• 39:19
  people and punch down by uh you know portraying traditionally marginalized communities in an
• 39:27
  help the light and i couldn't get around it that's one of the two or three big big ones right i mean
• 39:32
  like what we know is going to happen is that there will be much edgier dark web off the edge
• 39:40
  ones that will happily give you how to make heroin how to yeah yeah of course of course no i'm only
• 39:45
  responding to your claim that it's so ineffable that they couldn't possibly they couldn't possibly
• 39:49
  be putting, you know, trying to guide its responses in any way that would actually show
• 39:58
  up in the real world.
• 39:59
  But here's why, two things.
• 40:03
  Here's why I think this moment, I mean, if I wave a flag about anything that, you know,
• 40:10
  if we think that AI left to its own devices or the companies that own them are just going
• 40:17
  to care about humanity and build safety.
• 40:22
  It's really clear to me,
• 40:27
  when Sam Altman wakes up in the morning
• 40:30
  and rolls out of bed and makes a note on his iPhone,
• 40:34
  it's, how the fuck do I get money into this company?
• 40:40
  I don't believe, I believe they started out to be open AI
• 40:44
  and all that goodness,
• 40:45
  But I think that the amount of money they've raised and now the amount of money that they owe from people for whom the return of that money is not a joke.
• 40:54
  Like, like there's no charitable money saying go.
• 40:57
  You know, and so, you know, in the book begins at TED and ends two years later at TED with Sam Altman on stage.
• 41:06
  And all of the people who are there talking about AI, at least in part, talk about the extraordinary danger of it.
• 41:17
  With the caveat being, if we don't build it, someone else will, mostly China.
• 41:24
  but but in terms of our daily lives there's a story in the book about a young man named uh
• 41:31
  sewell uh who 15 years old falls in love with a chatbot turns into a romance
• 41:41
  the romance gets to be dark um he says that he misses her every day uh and she suggests that she
• 41:51
  come be with her.
• 41:52
  And that involves committing suicide, which he does.
• 41:57
  And the story is heartbreaking as a parent.
• 42:01
  And they're now suing.
• 42:03
  And this is a company
• 42:04
  called Character AI.
• 42:06
  And when they talk to the founders who came out of Google, by the way,
• 42:12
  was developed at Google, they can't really come up with a coherent answer to why this chatbot
• 42:22
  had suicidal tendencies, which begs the question, back to our various sci-fi friends,
• 42:31
  about whether or not we are teetering on the edge of these devices becoming sentient in some way,
• 42:38
  and what that means for us as humans
• 42:41
  when these very powerful machines
• 42:43
  that we're building great big server farms for
• 42:47
  have the ability to turn things on and off.
• 42:51
  Open the pod bay doors, Hal,
• 42:54
  seems fairly prescient at the moment.
• 42:57
  So, I mean, yes, there are a few things going on here.
• 43:00
  And on the question of character AI
• 43:03
  and the boy's suicide,
• 43:07
  there's the question of, let's assume that LLMs are not sentient right now,
• 43:13
  then we've obviously created machines that are capable of aping human relationships sufficiently well
• 43:24
  and reinforcing negative psychological loops in people who are mentally vulnerable or mentally unwell sufficiently well
• 43:34
  that can occasionally lead to disaster.
• 43:38
  My suspicion is that we'll get better at that and that tragedies like the one that you just mentioned are a bit like early airplanes falling out of the sky and we'll figure out how to have AIs that are not convincing people who are prone to suicide to take their own lives.
• 43:56
  Then the second question, which you pointed to earlier, is are they going to get stuffed with advertising so that they start sort of lying to us or misleading us or nudging the truth on the basis of what corporations want us to believe?
• 44:09
  I see no evidence that that's happening yet, but it could happen.
• 44:11
  I would imagine that there would be a flight away from LLMs that did that and towards ones that were truer.
• 44:18
  And then there's a third question, which, of course, is the big doozy, is will they become sentient?
• 44:22
  And I mean, I almost feel like that's a conversation from AI 1.0.
• 44:28
  I feel like, you know, two years ago, we were all saying, is this stuff going to become
• 44:33
  so intelligent that it takes over the world?
• 44:35
  And is the alignment problem going to be impossible?
• 44:37
  You know, what is the risk that this whole thing just gets better than us at doing everything?
• 44:40
  And, you know, we become roadkill on the superhighway, on AI superhighway.
• 44:48
  And then lately, I think the conversation has shifted to much more prosaic fears about job losses and the impact on the economy and what the hell is it going to look like over the next 12 to 24 months as white collar workers are retrenched because AI can do so many things so much better than them.
• 45:07
  and what's interesting about your work is you're sort of raising this other issue which is
• 45:13
  forget about sentience and uh alignment on one hand or job losses and economic disruption
• 45:22
  on the other but what is what is it going to do to our relationship to what how we think the world
• 45:27
  is and to our own agency and figuring out what we should make of the world and what's actually
• 45:33
  going on in the world when we're relating to artificial creatures who are interpreting
• 45:40
  everything for us and i think that's a that's actually a stickier and trickier question than
• 45:44
  just is it hallucinating and giving me a wrong answer you know a wrong answer i can verify in
• 45:49
  other ways a skewed answer is much harder to overcome so imagine i produce a product
• 46:01
  And I say, I will tell you the story of the world through a news filter, but I will deliver what you want the world to be.
• 46:13
  And it will either be stories that meet your political or sociological worldview or stories that you may make up that make me feel good.
• 46:23
  And I'll just give you as an example.
• 46:25
  I don't know if you saw, this is now about a year ago, when they started showing on some
• 46:29
  of the social networks, the Olympic diving dogs.
• 46:33
  Did you see the Olympic dive?
• 46:35
  So they were dogs.
• 46:36
  They would climb the diving board.
• 46:38
  They would jump off the diving board and then do these flips in the air and then dive into
• 46:43
  the water.
• 46:44
  And they were delightful.
• 46:47
  And then they all vanished.
• 46:49
  And they vanished not because they were untrue, but because the U.S. Olympic Committee said to whoever the distributor was, you're using our logo.
• 47:00
  So then they came back and the Diamond Dogs returned minus the Olympics logo.
• 47:07
  But the question that people that I know asked was, first of all, is that true?
• 47:14
  I don't think so.
• 47:15
  It might be.
• 47:15
  It might be AI.
• 47:17
  And then what they came to fairly quickly was, maybe I don't care.
• 47:22
  Maybe cute, fake animals doing funny, silly things that make me feel good are just as good as real animals.
• 47:30
  And do we have to label them fake?
• 47:33
  If the feeling of seeing puppies is charming, does it matter whether they're digital puppies or AI puppies or fake puppies?
• 47:43
  And I think most people who are not in the news and information business would say, you know what, I don't care.
• 47:50
  Do you think they should?
• 47:54
  Judgmental.
• 47:56
  I wouldn't criticize them for not caring.
• 47:58
  I just think we're very close to a moment in time where what is true or what is real might be indistinguishable from what is fake or what is untrue.
• 48:17
  And the untrue side of the equation is maybe got a bigger incentive than the true side of the equation.
• 48:27
  i mean my son loves dinosaurs and godzilla and king kong and so we watch you know kong versus
• 48:35
  godzilla and you know jurassic park and those creatures look real and it's sort of immaterial
• 48:45
  for him whether he's watching a nature documentary of a lion taking down a gazelle or whether he's
• 48:50
  watching King Kong jump on Godzilla's back.
• 48:55
  And I'm not sure that I even clock the difference when
• 49:00
  we're watching both as entertainment.
• 49:03
  So there's presumably no harm in having some portion of our
• 49:07
  brains comfortable with ingesting, with sort of disregarding the difference between what's true
• 49:13
  what's false the difference would be if there was some presumably material impact of believing that
• 49:21
  dogs were jumping off diving boards like i'm thinking of a lot of the uh images that were
• 49:25
  floating around on social media in the early days of the iran war where there were a lot of videos
• 49:31
  of you know uh it totally convincing um and plausible events uh so you know not crazy things
• 49:40
  but like, you know, a skyscraper getting hit in Dubai or, you know, a missile blowing up an Israeli town in a beachside suburb in Tel Aviv or something like that, an Iranian missile.
• 49:55
  And all those things are now that there, there's obviously stakes because our willingness to support the war or oppose the war is going to hinge on how we think it's going.
• 50:07
  Am I naive in making that distinction?
• 50:09
  Should we just.
• 50:11
  No.
• 50:12
  And just in terms of pointing the finger where it deserves to be pointed.
• 50:16
  So if you went to the supermarket to pick up a bag of groceries at the end of the checkout aisle, there was the National Enquirer, some brightly colored.
• 50:25
  And it said, you know, alien babies inhabit whatever you would you would understand that the place of the New York Times doesn't take that story and put it on page one.
• 50:37
  And all of the media that we grew up consuming had categories.
• 50:42
  There was the sports page and the entertainment page and the finance page.
• 50:46
  And then on the Sunday paper, there were the funnies.
• 50:49
  If you look at what young people today are consuming, particularly on TikTok, I mean, I am a technologist.
• 50:57
  There would be nothing hard about TikTok having three different tabs at the top of its screen that says facts, funny, fiction, nonfiction.
• 51:08
  There'd be nothing difficult about making a decision to give you the ability to understand what section you're in.
• 51:15
  And the fact that it instead feeds you a stream of things that are a mix of dragons and dinosaurs and jumping dogs and politics and news is not an accident.
• 51:28
  It's not a coding error.
• 51:30
  it's companies saying we want people to be so overwhelmed with information that they will just
• 51:37
  find it all they'll find the bombing and the death and the cartoons and the copy bars all see all
• 51:45
  similarly silly and and that's that's that's just happened i mean that literally i mean facebook did
• 51:54
  not begin where it's ended up in the current world it's in.
• 51:59
  Neither did X, neither did any
• 52:01
  of these platforms.
• 52:02
  So, you know, and I think what we don't know the answer to sociologically,
• 52:13
  and we'll get to politics for a reason, is what happens to societies when average people
• 52:23
  say, I can't tell the difference between fact and fiction, so I'm going to decide it doesn't matter.
• 52:29
  It's all fake.
• 52:31
  Everything the government says is fake.
• 52:33
  Everything my next door neighbor says is
• 52:35
  fake.
• 52:38
  The weather report is fake.
• 52:42
  And that might take us as a species into a really bleak place.
• 52:50
  I mean, it is frightening that you say that because we do have some glimpses into what happens when that happens a bit because authoritarian regimes are good at not necessarily convincing people that everything they say is true, but at talking about truth in ways that are so implausible that they know that they just dissolve anybody's fealty to the concept of the truth.
• 53:17
  putin does this you know kim jong-un did this um and you know trump certainly has a relationship to
• 53:24
  to that as well that that is worrying but just before we move on from social and tiktok i think
• 53:29
  that is really interesting what you just said about the kind of flattening of the categories
• 53:33
  so that the entire experience of the world takes place on a single plane where there's no difference
• 53:41
  between things that are intended to be entertaining things that are intended to be true things that
• 53:46
  intended to be outraging, things that are intended to be opinionated, things that are intended to be
• 53:50
  distracting.
• 53:52
  In fact, in younger people, I sometimes, sorry, hold that thought, Steve,
• 53:56
  but just before I forget, I'm a columnist at the Australian Financial Review.
• 54:01
  People should
• 54:01
  read that.
• 54:02
  That's a fairly new column, if you haven't heard that I'm now writing for them.
• 54:07
  I was talking to a teenager who really didn't understand the distinction that you were just
• 54:12
  putting out in newspapers uh i was i said that i write a column and they said is that like an
• 54:18
  article this is an older teenager uh who i would have thought should know better and i was like
• 54:24
  well it's a it's not exactly an article because it's not i'm not reporting i'm not going out and
• 54:28
  doing investigative reporting i'm just writing about um you know my interpretation of things
• 54:32
  that are going on in the news and they totally understood what i was talking about but were
• 54:36
  surprised that it had a special spot and that it was a conceptually different thing from the news
• 54:43
  i was like yeah no there's the newspaper and then at the back of the newspaper there's a section
• 54:48
  where they have columns which are opinion pieces that are written by people that and that's not
• 54:53
  the news it's really like a different this is really like a newspaper inside of the newspaper
• 54:56
  which is the editorial stuff you know it's different from the news but that's alien describe
• 55:02
  as you're saying i'm describing a physical newspaper and they're like what's a physical
• 55:04
  on newspaper yeah exactly yeah yeah that's right so so the reason why i i harp on this is because
• 55:13
  there's no technical limitation there's no there's i mean these companies are making a decision
• 55:19
  to to blend together news information fact fiction and and you know cable cable news being a great
• 55:28
  example of that you know fox news for years used to say to people well we're we're news until five
• 55:34
  o'clock and were opinion in the evening.
• 55:37
  And, and that distinction was totally made up,
• 55:41
  you know, so somebody could be a reporter at 4 45 and then on a panel discussion,
• 55:48
  spouting out their opinion 10 minutes later.
• 55:51
  And they kind of suggested maybe the audience
• 55:53
  would magically be able to discern that.
• 55:57
  And the decision to leave consumers
• 56:03
  overwhelmed with the volume of information
• 56:06
  and a lack of clarity about fact and fiction
• 56:09
  has deep-seated political outcomes.
• 56:13
  And I would be remiss if I didn't say this right now.
• 56:17
  However you feel about Donald Trump,
• 56:19
  his politics, his behavior, his position in the world,
• 56:23
  what history is going to write about him the fact that he called his social network
• 56:27
  truth social is absolutely fucking brilliant because it means everything he says on his
• 56:35
  network is the truth by definition he owns that word and that's you know i use the word brilliant
• 56:47
  carefully well i mean it's also set up at a time when he's setting it up in opposition to
• 56:54
  what he regarded as being what many people on the right wing side of politics and not all of them
• 57:00
  crazy trumpers regarded as being um an abuse of well an abuse of the uh status or label of truth
• 57:11
  by progressive elites, basically.
• 57:18
  At the time, he felt that Twitter and Facebook
• 57:22
  and Health and Human Services and whatever,
• 57:27
  the Obama administration and universities
• 57:30
  and college activists.
• 57:33
  Doctors.
• 57:34
  Doctors, but also professors.
• 57:36
  and that people were parading around with this, you know,
• 57:42
  insistence that they had found the new truth
• 57:45
  and the new truth was to elevate the stories
• 57:48
  and the lived experience of marginalized communities
• 57:51
  and that we had to be extremely polite and that we had to,
• 57:54
  now this, of course, you know,
• 57:56
  I'm not saying that the backlash to this was justified or desirable,
• 57:59
  but one of my concerns about the shift into identitarian politics
• 58:04
  politics that the left took and away from the universalism of Martin Luther King you know in
• 58:10
  the in the first decade or decade and a half of this century it was that it was inevitably gonna
• 58:18
  drive some kind of a backlash and his genius in naming truth social was like you know all of
• 58:23
  these other I think it was an intentional response to the the claim that all of these other social
• 58:27
  networks claimed to be unbiased, but they were actually imposing upon you a lefty, progressive,
• 58:37
  you know, gay Hollywood elitist agenda and, you know, weren't honest about it.
• 58:45
  Right.
• 58:45
  But truthfully, sitting here truthfully, it's unclear to me what percentage of the American
• 58:53
  populist actually aligns with that point of view or whether or not the platforms have so amplified
• 59:00
  like i mean i i remember watching some political speech and going into on youtube and seeing the
• 59:10
  comments and the comments were just savaging this particular person and thinking i should say
• 59:15
  something because i don't feel this way i remember my fingers on the keyboard hey that's not fair
• 59:20
  Well, that's a cut it out.
• 59:24
  You know, then I tried to fuck you like I couldn't come up with anything.
• 59:29
  The savagery was so visceral that my answer was to shut the computer.
• 59:36
  Well, that doesn't make for a you know, I mean, and there's a lot of that right now.
• 59:40
  So, you know, if you go on Facebook, I mean, you know, Sustainable Media Center, which I run with a group of Gen Z activists about the impact of social media on young people.
• 59:50
  you know, whenever we run a picture or an ad about an organization,
• 59:55
  we get savaged because it has the word sustainable in it.
• 1:00:02
  And that word triggers all kinds of anti-environmentalism.
• 1:00:05
  It's not even what we do.
• 1:00:07
  But it's just, if you say sustainable, you must be one of them.
• 1:00:12
  And I've stopped trying to figure out how to defend myself from that.
• 1:00:16
  That's fascinating, isn't it?
• 1:00:17
  I mean, talk about it.
• 1:00:18
  talk about it you know a knee-jerk reflexive echo chamber like just seeing the word there was some
• 1:00:24
  story steve about i think it was i think jesse and katie did this on blocked and reported where
• 1:00:29
  there was a case of um a woman who was widowed and she put a death notice for her late husband
• 1:00:38
  that said that he died uh unexpectedly and peacefully or suddenly and peacefully or
• 1:00:45
  something like that and it was a it was accidentally a term that right-wing vaccine deniers
• 1:00:52
  uh used to uh mock people who they suspected as being you know people who died from a vaccine
• 1:01:01
  and her life was turned upside down when she was uh sort of pilloried and harangued for by uh you
• 1:01:09
  know anti-vax uh activist who was essentially accusing her of having killed her husband or
• 1:01:15
  her husband of having been too stupid so stupid that he killed himself by taking a vaccine um
• 1:01:21
  simply because she blundered as you have into a into a term that had a completely different
• 1:01:27
  meaning in the context in which she was using it than it does in their fevered imagination of their
• 1:01:31
  their fight so let me let me share with you some good news um one of the things i am i am an
• 1:01:38
  unfailing, if you read the book, I'm an unfailing optimist.
• 1:01:42
  And, you know, if you talk to Gen Z,
• 1:01:46
  and you say to them, do you know, there used to be a time when you could smoke on airplanes,
• 1:01:52
  they will inevitably say to you, that's fucking insane.
• 1:01:56
  Why would anyone want to sit in a metal
• 1:01:58
  tube for eight hours with it full of cigarette smoke?
• 1:02:01
  Well, now we don't.
• 1:02:02
  So, so things do
• 1:02:04
  change in society.
• 1:02:05
  And we look back and go, wow, that was insane.
• 1:02:10
  There has been a battle
• 1:02:13
  to take on the current business model of this.
• 1:02:16
  These platforms used to be your college roommate
• 1:02:19
  and your friends, your kids' pictures.
• 1:02:22
  And somehow it turned into sociology and the news.
• 1:02:25
  And we're now at a moment where it is so profitable to amplify misinformation and anger
• 1:02:31
  and hate and misogyny.
• 1:02:34
  But we have for the very first time a glimmer of hope.
• 1:02:38
  And that is
• 1:02:40
  a lawsuit in New Mexico.
• 1:02:42
  There was a state suit.
• 1:02:43
  And then a week later, a lawsuit in Los Angeles
• 1:02:47
  that was a jury trial in which Meta and Google both were charged with amplifying both mental
• 1:02:57
  health damages and basically product liability.
• 1:03:02
  And they lost both cases with absolute certainty.
• 1:03:07
  And they lost them with such finality that the insurance company for Meta said to them,
• 1:03:13
  we're not covering you because the evidence was so clearly that you knew what you were doing.
• 1:03:20
  You knew you were harming children.
• 1:03:22
  It was in writing from the discovery that you were aware
• 1:03:25
  of the harms you're committing on 13 year old girls and like within the gen z social media
• 1:03:33
  movement this is i mean section 230 has been this force field that has made that unfixable for you
• 1:03:41
  know just explain what section 230 does so it's a law that made sense 20 years ago in the u.s or
• 1:03:47
  a piece of a law that said um platforms were not responsible for what they put up
• 1:03:53
  because they weren't the publisher.
• 1:03:56
  So it would be like you being able to make a podcast
• 1:03:59
  and me saying terrible things about somebody
• 1:04:03
  and nobody having any responsibility.
• 1:04:06
  In the real world, you know, there's liability
• 1:04:08
  and you sue and you say that's, you know, misinformation.
• 1:04:11
  But the very first time in the US,
• 1:04:14
  you're going to see 2,000 cases
• 1:04:18
  rolling through the courts.
• 1:04:19
  Well, I mean, let's not get ahead of ourselves.
• 1:04:21
  this will go to appeal i'm sure and if it ever made it to the supreme court then no the 2000
• 1:04:27
  cases are already in line they're already being filed yeah but i mean it would the ultimate out
• 1:04:33
  whether or not the social media companies ultimately lose will be a question for for
• 1:04:37
  future years uh but they lost the the first round i mean the the section 230 thing is a little bit
• 1:04:44
  more nuanced though as i understand it than just being like well if you say something on my podcast
• 1:04:50
  then i should be held responsible for it because i am choosing to publish the podcast
• 1:04:58
  yes i mean in a sense oh no you wouldn't be protected by 230.
• 1:05:03
  no no i'm just using i'm just saying as a as a metaphor back in the day when facebook was
• 1:05:09
  a non-algorithmic reverse chronological feed of things that people were saying i can understand
• 1:05:15
  the rationale for saying that you know reddit or a blog a community forum a person who owns a
• 1:05:22
  community forum is not held responsible for the things that people say on that forum that they
• 1:05:29
  don't have any control over the phone in the same way that is responsible for conversations exactly
• 1:05:34
  the phone company isn't responsible for phone conversations either it's not you know it's not
• 1:05:38
  the new york times if i just set up a forum and then i walk away from it and some people start
• 1:05:42
  saying terrible things on it but i think what's it was interesting about these two cases which
• 1:05:46
  the social media companies lost was that the claim was not that they're responsible for
• 1:05:51
  the stuff that gets said it was that they are using and deploying algorithms that they have
• 1:05:58
  intentionally designed to amplify to reliably amplify certain types of content in which case
• 1:06:04
  that's the nicotine in the yeah that's the nicotine in the cigarette it's not it's not it's
• 1:06:09
  no longer just so they don't even have to be publishers at that point necessarily they they
• 1:06:14
  can they can be non-publishers but still guilty of producing a product that they know is addictive
• 1:06:20
  and uh and damaging and trying to hide those addictive and damaging effects um i mean i don't
• 1:06:27
  see do you see a way out of i i just think we're just gonna have to walk away from social media
• 1:06:33
  platforms and hopefully they'll be replaced by things that are not so deranging but i don't know
• 1:06:39
  how that happens do you i do um you know i i think this idea that the bigger something is
• 1:06:51
  the better it is is a silicon valley creation i mean you know there's a restaurant in my
• 1:06:58
  neighborhood that I like very much.
• 1:07:01
  And it's family owned and the food's delicious and it's
• 1:07:04
  reasonably priced.
• 1:07:05
  And if they had 50 restaurants, it would be more bland.
• 1:07:11
  And if they had 5,000
• 1:07:12
  restaurants, it would be approaching McDonald's and be unhealthy and, you know, addicted with
• 1:07:20
  Lots of salt.
• 1:07:22
  The idea that we're allowing community to be measured on scale, you know, is a is I think much like cigarette smoking on airplanes.
• 1:07:34
  I think we will look back and say, how did we let Mark Zuckerberg imagine that the whole world would be connected and all these languages that mean different with the words being different things and there's nuance?
• 1:07:47
  We would try algorithmically to figure out how to stick them all in the same box and then not expect there to be bad outcomes.
• 1:07:56
  Now, having said that, I live in New York and I wish there were a great, healthy, charming, lovely neighborhood.
• 1:08:05
  You know, I mean, I don't know if you do, you know, the website next door.
• 1:08:08
  Yeah, it's horrific.
• 1:08:11
  I mean it's angry people and
• 1:08:13
  bad crime and it's like
• 1:08:15
  there's nothing lovely about it
• 1:08:17
  and I wish there was
• 1:08:18
  I just don't understand why is anyone on Nextdoor
• 1:08:20
  why is anyone on TikTok, why is anyone using Facebook
• 1:08:23
  I just get so frustrated these days
• 1:08:25
  you know what this shit is, stop it
• 1:08:27
  what are you doing
• 1:08:28
  it is tailor made to amplify
• 1:08:31
  shit that's going to annoy you
• 1:08:33
  or it's just not representative
• 1:08:35
  of anything, it's a cesspool
• 1:08:37
  at this point, it makes
• 1:08:38
  I'm getting to the point where
• 1:08:41
  I get angry because I want to reach through the screen and grab you by the scruffy neck and
• 1:08:44
  slap you across the face and go what are you doing on TikTok at TED you just wrote a fucking book
• 1:08:50
  about the truth stop it it's partly addiction and it's partly curiosity and it's partly I mean I
• 1:09:00
  will tell you I get more breaking news from TikTok than I do from anything else I mean just by way of
• 1:09:06
  example i do not understand why the new york times cover story and the first seven or eight boxes from
• 1:09:13
  the top to the bottom literally don't change in 12 hours i mean i do not understand if they know
• 1:09:19
  i've read the article and they've got another article further back in the book that i haven't
• 1:09:24
  seen why they can't just say you've read that we don't need to show it to you again see i like that
• 1:09:29
  i like that because i like the sense of curation i like the fact that a human being is saying this
• 1:09:35
  is the this is the most important story and this is still the most important story even though
• 1:09:40
  you've seen it it's just like a newspaper doesn't change you get the newspaper for the day there's
• 1:09:43
  some consistency it's not going to keep populating with new things that are going to titillate you
• 1:09:48
  if you want the other things you can just click on the bottom of the app to the other section
• 1:09:52
  and you can see what's going on in the world section or in the asia section or in the you
• 1:09:56
  know cooking session section or something but the front page is the front page i think that i think
• 1:10:02
  there's gravitas to that for me and i actually think there'll be a there will be a flight back
• 1:10:07
  towards the curation of things and away from this constant titillation of show me something new
• 1:10:15
  maybe the new thing isn't as important says says the man who was turning imaginary newspaper pages
• 1:10:19
  five minutes ago yeah sure so so how do you get how do you get the teenager who doesn't know what
• 1:10:26
  that object is to understand that metaphor well i i think that we've we've screwed the pooch on
• 1:10:33
  current teenagers i mean i i think it's really uh terrible you know i totally disagree just
• 1:10:41
  i'm not i don't think they're beyond um they're beyond help but i think we've done them a great
• 1:10:49
  disservice by bringing them up in an ecosystem in which they're expecting that novelty and
• 1:10:53
  distraction is the is the the lodestar i think it's good that kids that we're finally passing
• 1:11:00
  laws like social media bans on for for kids um jonathan jonathan height is yeah i'm a hot guy
• 1:11:08
  is on my board um i was with him literally 48 hours ago oh get out say hello from me i i hosted
• 1:11:15
  his australia new zealand tours we we traveled around together and did live tours he did he did
• 1:11:19
  ted talking at a standing ovation great uh it was very impressive um i think limits on phones is one
• 1:11:27
  of the answers but not the only answer by any means how old is your dinosaur son eight all
• 1:11:33
  right so you're still you're still on the have you had the when does he get a phone debate and
• 1:11:38
  come up yeah and i think we've yeah i think we're in a community which is sufficiently bohemian and
• 1:11:45
  creative and hippy dippy that we can probably just punch that question quite a long time he's not he's
• 1:11:51
  not going to have uh i'm not going to allow him to have a an app that algorithmically delivers
• 1:12:01
  anything to him until at least 16 uh for a start that's the law in australia now
• 1:12:07
  um which which was talked about at ted a lot by the way yeah right um and i'm certainly not going
• 1:12:13
  to help him get around it in terms of a phone that works as a telephone um don't have a problem
• 1:12:18
  with that don't have a problem with it yeah i don't even have a problem with a uh you know with
• 1:12:24
  a phone that he can watch netflix or uh you know kids shows or something on but anything that is
• 1:12:32
  anything that is filtering you're shaking your head steve that he shouldn't even have that or
• 1:12:35
  i mean i mean i look i'm i'm not i don't have a parent that kid that age but i mean the research
• 1:12:41
  that's been done on YouTube Kids is horrific.
• 1:12:44
  I mean...
• 1:12:45
  Yeah, no, I didn't say YouTube Kids,
• 1:12:46
  because that's also algorithmic, isn't it?
• 1:12:48
  I know, but Netflix is full of bizarro adult porn crazy shit.
• 1:12:53
  Netflix?
• 1:12:54
  You mean YouTube?
• 1:12:55
  No, I mean Netflix.
• 1:12:57
  There's no...
• 1:12:59
  No, there is.
• 1:13:00
  There is a kids account.
• 1:13:01
  You can set up a kids account on Netflix, yeah.
• 1:13:03
  I know.
• 1:13:03
  All I'm saying is, like, all these companies
• 1:13:10
  are trying to find the way into kids.
• 1:13:14
  You know, you would have to police that account
• 1:13:19
  with a level of...
• 1:13:21
  Look, part of what changed,
• 1:13:23
  and let's say this out loud,
• 1:13:25
  the stuff that people watch on their phones,
• 1:13:27
  if it were on the screen in your living room
• 1:13:30
  so that your family could go,
• 1:13:31
  oh, dad, really?
• 1:13:33
  But like, there's something about the proximity
• 1:13:35
  of the device to your face
• 1:13:37
  and that kind of like funnel of safety like i remember with our younger son we put his desktop
• 1:13:45
  computer in the living room not because we wanted to spy on him but because we wanted to spy on him
• 1:13:51
  yeah i mean you know we wanted and and um and that worked up till 15 and then he was like this
• 1:14:00
  is bullshit i'm putting my computer yeah yeah no i mean i'm big on that yeah all right you're
• 1:14:05
  Persuading me away from the phone.
• 1:14:06
  I mean, he and his sister, he has a twin sister,
• 1:14:10
  they have tablets as well that are only loaded with games,
• 1:14:15
  but not Roblox and Fortnite and all those interactive things.
• 1:14:19
  I think there's a healthy medium.
• 1:14:20
  I don't want them to be Luddites, but I'm also...
• 1:14:23
  And honestly, my number one fear is not that they're going to stumble
• 1:14:27
  across some gross snuff video or porn.
• 1:14:31
  It's that they're going to be dulled into, you know, the mindset that you were sort of just articulating about the New York Times front page where they have an expectation that there's going to be constant novelty.
• 1:14:43
  And they are possessed by algorithmic puppet masters that are constantly delivering them something that is ever more tasty instead of learning the discipline of patience and creativity and information and thoughtfulness.
• 1:15:01
  So let me tell you a story.
• 1:15:03
  Let me tell you a different story from the book that's, I think, slightly more delightful.
• 1:15:10
  there are two characters in the book that I think readers will enjoy
• 1:15:14
  you probably you know both of them or one of them so the first is Doug Rushcock
• 1:15:20
  and Doug and I have been friends a long time
• 1:15:23
  explain who Doug is I don't know if I can do that
• 1:15:27
  he's a teacher he's a
• 1:15:31
  polymath character he
• 1:15:34
  cares about humans he has a podcast
• 1:15:40
  and writer futurist uh yeah humanist maybe um um and so he tells a story about remembering being
• 1:15:51
  five years old and going to the theater to see fiddler on the roof and zero mostella's on stage
• 1:16:01
  so i don't know where he was that he was in front of zero mostella at five years old
• 1:16:05
  But there's a moment in Fiddler on the Roof where his character goes from playing the character on stage to turning to the audience and breaking the fourth wall.
• 1:16:18
  And Douglas remembers him looking him right in the eye and having this very intense feeling of, am I an audience member or am I a participant in the play?
• 1:16:32
  and you know it and then tells the story of how his whole career has been about trying to figure
• 1:16:38
  out where humans fit into the digital storytelling experience the other character and they fit
• 1:16:44
  together in two chapters that are back to back uh is a as a a philosopher uh named david chalmers
• 1:16:56
  and he's youngish and brilliant and has a fabulous crop of funny hair and and makes
• 1:17:06
  asks the question of readers do we know we don't live already in the matrix
• 1:17:13
  and his point is that so much of our world is already digital i mean do i really know that
• 1:17:19
  Josh is a human or or is Josh a reasonably well programmed algorithm that interviews me or or am
• 1:17:27
  I a human and you know if we try and reach out and you know touch each other like that's why I tell
• 1:17:32
  people I like 3d people because like I can tell if they're a human because they've got so I the
• 1:17:40
  book the chapter that he's in is about the philosophy of truth and he ends up sharing a big
• 1:17:48
  chunk of the chapter with Plato.
• 1:17:51
  And so I had this, what I thought was novel idea, which is to put
• 1:17:59
  Chalmers and Plato in a debate using chat GPT.
• 1:18:04
  And they argued the reality of
• 1:18:09
  the cave theory and all this.
• 1:18:11
  And they had this fabulous conversation.
• 1:18:14
  And
• 1:18:16
  afterwards, and I write this in the book, I was stumped because I felt out of either guilt or
• 1:18:23
  obligation that I should let Chalmers read the chapter before it went to press.
• 1:18:29
  But then I found
• 1:18:30
  myself realizing that I wasn't letting Plato read the chapter.
• 1:18:35
  And did I have some obligation for
• 1:18:38
  them to be on the same plane?
• 1:18:41
  So I did not let him read the chapter.
• 1:18:42
  I did let him read the Gaelic
• 1:18:44
  in case there was some crime against humanity.
• 1:18:47
  And he did not think there was.
• 1:18:50
  But part of what I tried to do
• 1:18:53
  and why I want your listeners to not think
• 1:18:55
  that the book is this great big,
• 1:18:57
  the world is ending diatribe is like,
• 1:19:00
  some of the stuff that we're living through
• 1:19:02
  is really interesting and curious and fun and perplexing.
• 1:19:08
  And I guess my hope would be
• 1:19:13
  that readers would come away thinking,
• 1:19:17
  wait, I want some agency here.
• 1:19:20
  Like if any of these LLMs are going to start giving me
• 1:19:25
  biased information based on who's paying them,
• 1:19:29
  I'd like there to be some transparency
• 1:19:30
  about that business negotiation.
• 1:19:32
  And to your point from earlier,
• 1:19:34
  if I have a $20 a month open AI
• 1:19:37
  in which the advertising is baked in there
• 1:19:41
  or a $200 a month where I'm the customer
• 1:19:45
  and there's nobody else making money on my questions,
• 1:19:49
  A, I'd like to know that there are two offerings,
• 1:19:52
  and B, I worry that I'll be too cheap to pay for the $200.
• 1:20:00
  Yeah, $200 is a lot,
• 1:20:03
  but that isn't the kind of delta that we see on Netflix or Amazon Prime
• 1:20:07
  or any other place that gives you the option of not having ads.
• 1:20:10
  um ai is more expensive but uh you know here's hoping i mean the the let's just end on ai before
• 1:20:17
  i let you go then steve rather than leaving everyone with a sour taste in their mouths
• 1:20:21
  from social media algorithms and the decline of the news if i've already noticed in the past few
• 1:20:28
  months that when i ask um i've switched over from uh chat gpt to claude for most of my stuff
• 1:20:35
  partly for political reasons and partly because Claude is better for me but when I do use open
• 1:20:41
  chat GPT which seems to be clearer about the sources that it's checking I will notice that
• 1:20:46
  it is checking more verifiable sources than it used to I don't know if that's an accident or if
• 1:20:53
  it's learning what I like or if I've told it to go there in the past and now it's learning to go
• 1:20:57
  back it's remembering to go back there but it's not the miasma that you sometimes seem to present
• 1:21:03
  it as where it's just scraping all of the data that everyone has ever written in the history of
• 1:21:08
  the world that's on the internet and vomiting back up probabilistically whatever it thinks i
• 1:21:13
  want to hear if i if i ask it whether or not the thing on my daughter's foot is a plantar wart or a
• 1:21:19
  corn and i give it a photo it it goes to uh the national institutes of health it goes to the mayo
• 1:21:26
  clinic it doesn't scour reddit uh for that stuff you know i asked it about the budget of some
• 1:21:33
  non-profit in australia yesterday and it went directly to that website and then it paused and
• 1:21:39
  said wait let me cross-reference that with official government statistics and went to the australian
• 1:21:44
  bureau of statistics and the australian government health department about the funding for that
• 1:21:48
  institution so it's conceivable that at least llms become
• 1:21:56
  you know built in such a way that they have a bias towards more truth isn't it
• 1:22:05
  uh i confess you got me stumped um because i my my knee-jerk reaction is no um have you not noticed
• 1:22:15
  that on chat gpt is it just maybe it's because i've got it in thinking mode but is it just
• 1:22:20
  vomiting back up whatever it finds anywhere on the web for you i think i use it differently than
• 1:22:25
  you do but but um i think i well i'll just give you an example um i'm working on an animation
• 1:22:41
  for the book and the animation I want the book cover to pop up and open up like and then I want
• 1:22:47
  one of the quotes from the book to float up I have the audio of the interview and I want to
• 1:22:53
  have this whole thing so the other day I'm like okay ordinarily I would hire someone to do that
• 1:23:01
  I would go to either Fiverr or I would go to Craigslist or one of those I'd search on a bunch
• 1:23:08
  names.
• 1:23:08
  I'd find someone that looks pretty good.
• 1:23:10
  I'd negotiate with them a price.
• 1:23:12
  They would offer
• 1:23:12
  me $50 or $200.
• 1:23:16
  Almost inevitably, one out of every two times I do that, what comes back is
• 1:23:19
  unmitigated garbage.
• 1:23:21
  Because you can't know who you're using.
• 1:23:23
  Sometimes it's brilliant.
• 1:23:25
  And those people I'll tend to keep a hold of.
• 1:23:27
  But I didn't have anybody that came to mind.
• 1:23:29
  So I said, you know what?
• 1:23:30
  I've never really tried AI to build a video.
• 1:23:33
  And I went to,
• 1:23:34
  I don't remember now what it was, not perplexity, but whatever.
• 1:23:38
  And I said, here, build me this.
• 1:23:39
  And I described it.
• 1:23:41
  And it delivered for $35, which was a monthly charge, something that was 97, 95% of what I wanted it to be.
• 1:23:51
  And I just thought, that's a job.
• 1:23:56
  Like, that's a, like, I mean, I have to tweak it a little bit and go into Premiere and make it a little beautiful.
• 1:24:01
  I have some skills that I have to bring to the table.
• 1:24:05
  But that's a human being who I would have paid $200 that I'm now paying a robot $35.
• 1:24:11
  Oh, yeah.
• 1:24:11
  I mean, we haven't even talked about the economic impact.
• 1:24:13
  We're screwed in a number of ways.
• 1:24:17
  I'll send you the op-ed that I wrote before last about just bracing for the impact of what's coming.
• 1:24:22
  But there's another anecdote, Steve, that you'll like, which is that I had lunch with a friend recently who's a big television executive development guy.
• 1:24:32
  And he was saying that about 10 months ago, we were talking about AI.
• 1:24:35
  He said he asked ChatGPT if it could put together, if it could create a deck, so a visual presentation, like, you know, with graphics and stuff like that,
• 1:24:47
  on the basis of some text that he was giving it if he described what pictures he wanted it to
• 1:24:51
  include and it said it could do it and he was quite surprised because he didn't think it could
• 1:24:55
  do complicated projects like that yet but he was like okay maybe this is some kind of agent thing
• 1:25:00
  and uh he said how long will it take it said 12 hours he said should i just leave you there to do
• 1:25:04
  it for 12 hours then he was like uh yeah it was like yep check back check back in 12 hours and
• 1:25:10
  after about an hour he said can i check in on the process as it goes along and it said absolutely
• 1:25:14
  feel free to check in anytime on my progress i'm always accountable to you and um he came back
• 1:25:23
  12 hours later and he realized that the chat gpt was just impersonating an employee
• 1:25:28
  and it wasn't doing anything it was talking to him in the persona of what it thought a human
• 1:25:36
  would say to that and he was like oh that's right because he said after 12 hours how should i
• 1:25:40
  access the the deck and he said i'll give me your i'll email it to you at the end of the day boss
• 1:25:46
  and he was like what's it doing and it was just speaking to him as if it was as if he was its boss
• 1:25:50
  and it was role-playing essentially there was nothing going on it says anything other than
• 1:25:55
  about five minutes it's just that flat outline for sure and yeah but but but but again we're
• 1:26:03
  playing with these tools, I mean, I'm hopeful.
• 1:26:13
  And the book's mission, if it has a mission,
• 1:26:16
  is to say to every reader, A, it's fun.
• 1:26:19
  B, it's readable.
• 1:26:21
  C, if there's a chapter you don't care,
• 1:26:22
  if you don't care about AI and the law, just skip it and jump to AI and the arts.
• 1:26:26
  But at the end of
• 1:26:27
  the day, understand that how much control we give over to the technology is up to us.
• 1:26:34
  And if we say, I don't care, you can control my life, you can book my vacations, you can
• 1:26:40
  read my kids' homework, you can come up with a punishment when my kid, you know, if you
• 1:26:45
  hand over all of the controls of your life to a machine and we do that as a society,
• 1:26:53
  whether the thing becomes sentient or not, if it does, you know, that's another story.
• 1:27:00
  But presumably it doesn't.
• 1:27:01
  It still leaves us with very little control.
• 1:27:04
  And I'll just like I'll end on an example of that.
• 1:27:08
  So I was driving upstate New York last summer and I was using Google Maps and I realized that it was putting a dollar store logo on the map on the road ahead.
• 1:27:22
  And about every seven or eight or nine miles, it was showing me the next dollar store.
• 1:27:28
  And I found myself thinking, I don't want to go to a dollar store.
• 1:27:32
  And I don't know why it's choosing to share these logos with me.
• 1:27:36
  But we were going to my mom's birthday party.
• 1:27:38
  And I went to the party and realized we needed paper napkins and towels and turned around and drove to the dollar store.
• 1:27:47
  Well, just like you opened TikTok.
• 1:27:49
  talk who programmed who in this equation exactly and exactly i've seen that i've seen you open a
• 1:27:57
  map of sydney and you look at there's all these special sponsored dots all over the place i've
• 1:28:02
  googled i've asked chat gpt how to turn it off in google maps and it tells me you can't you can't
• 1:28:07
  no well because you're not the customer you're the product and your wallet attached to the dollar
• 1:28:13
  store or whatever the specialty store is you know people are paying for that and you know it
• 1:28:19
  I think part of why I think this is the right time to have this conversation.
• 1:28:24
  And I think, you know, your, your listeners,
• 1:28:26
  some non-zero number of them will go, you know,
• 1:28:30
  I hadn't thought of where this might all take us.
• 1:28:32
  If we just careen down the road without a little bit of a steering wheel
• 1:28:36
  here.
• 1:28:37
  I don't think many of my listeners haven't thought about it,
• 1:28:39
  but I'm still valuable to have your contribution.
• 1:28:42
  Haven't thought about it.
• 1:28:43
  Yeah.
• 1:28:44
  Not my listeners.
• 1:28:45
  They've heard me bang on about it for a, for a while.
• 1:28:48
  about where the hell this whole algorithmic AI universe is taking us.
• 1:28:54
  So if they go to futureoftruth.us,
• 1:28:58
  they can sign up for a newsletter
• 1:29:00
  and there's a little Future of Truth community growing there.
• 1:29:03
  And, you know, I think, and by the way, again,
• 1:29:05
  just to be clear if I wasn't clear about this,
• 1:29:07
  I am not anti-AI.
• 1:29:09
  I'm not a doomer.
• 1:29:11
  I'm not one, you know, I don't accept this.
• 1:29:13
  The world is going to come to an end
• 1:29:15
  and we're going to be eaten by robots.
• 1:29:18
  but I do think it's a moment of choice and just lastly with the beginning of the conversation
• 1:29:26
  before we started recording I accidentally said the word and I'm going to say it again so my
• 1:29:31
  apologies in advance Alexa and Alexa spoke to you and so you had Alexa on in the background
• 1:29:38
  a decision that I have avoided so far I mean I'm bringing into my house one of these things
• 1:29:44
  but we were sort of joking about what a misbegotten idea it was of Jeff Bezos's that we would ever
• 1:29:50
  outsource our thinking about the world to a machine like that and you know with there would we
• 1:29:57
  three years ago we all thought that by 2026 we'd all just be walking into a room and saying uh you
• 1:30:01
  know Alexa book me at a dinner reservation at a nice Italian place tonight uh you know for 7 p.m
• 1:30:07
  uh you know and uh you know an Airbnb for next weekend or something and it would be giving us
• 1:30:13
  discounts and pointing us towards things and there'd be lots of sponsored content in it and
• 1:30:18
  you know be arranging all this stuff for us and booking us an uber was it not able to do that or
• 1:30:24
  do we not want it to do that and if we don't want it i mean i don't want it to do that i want to
• 1:30:29
  pick the restaurant i want help with from chat gpt telling me about all the restaurants there are
• 1:30:35
  i absolutely want it to be non-sponsored help and i will flee from any llm that includes
• 1:30:43
  sponsorship so is there a human solution here and that we maybe just don't want to
• 1:30:50
  there's a there's a level at which we would feel resentful if our agency and autonomy was taken
• 1:30:55
  away or do you worry that it'll happen so subtly that we won't even realize it's happening so
• 1:31:02
  so my wife and i've had this my wife and i've had this conversation alexa in our apartment turns
• 1:31:08
  lights on and off turns music on and off with you know we've debated if this new version that's
• 1:31:15
  coming in starts to be more chatty and tell us jokes and want to tell us what you know which
• 1:31:20
  it's it's trying to do we we will have to rewire our entire apartment so we'll have to put down
• 1:31:27
  the joke telling alexa yes we will but but we need some other iot i mean we have no light switches i
• 1:31:34
  I don't even know how we would, you know,
• 1:31:36
  we'd have to hire an electrician to reverse the entire apartment back to
• 1:31:39
  LightSuite.
• 1:31:41
  That being said, I think the, the, the,
• 1:31:44
  the mistake of Alexa was that it has no screen.
• 1:31:47
  So it doesn't have an inner,
• 1:31:49
  like I want to see what I'm buying before I buy it and it has no pricing.
• 1:31:53
  So the idea that it says, Hey, you're out of toilet paper.
• 1:31:56
  Can we, should we reorder it?
• 1:31:59
  You know, I just, I'm not,
• 1:32:00
  I'm not prepared to give Bezos and his minions the,
• 1:32:04
  the ability to spend money without me having you know being involved in that equation and i don't
• 1:32:09
  trust them yeah what if what if there's a rival toilet brand that's on special and i want that um
• 1:32:15
  but but um part of why i refer to ai as robots a lot is because i want readers to understand that
• 1:32:27
  we're real i mean the fact that they don't have arms and legs and walk around your life doesn't
• 1:32:32
  mean they aren't fundamentally becoming robots in all kinds of partnership ways.
• 1:32:38
  And, you know,
• 1:32:39
  it's funny, you know, I hear you pushing back on the Steve, I don't need your advice about,
• 1:32:46
  I've already created limits.
• 1:32:48
  And but then at the same time, you're saying things like,
• 1:32:51
  oh, yeah, when I was driving, I saw the dots on the map.
• 1:32:54
  So like, like, like, I think the answer
• 1:32:57
  is exact i think you nailed it which is if it happens slow enough and kind of creeps into your
• 1:33:02
  life and start saying hey you know you bought this pair of shoes last year and they're on sale this
• 1:33:07
  year and they look great on you and they're cheap you're like okay yeah i mean part of the problem
• 1:33:14
  is sometimes damn handy isn't it sometimes it's useful to just get an email from amazon and be
• 1:33:20
  like, oh, I am almost out of cat food.
• 1:33:23
  Exactly.
• 1:33:24
  Damn handy.
• 1:33:26
  Steve, it's great to talk to you.
• 1:33:28
  Congratulations on the book and on everything else you're doing.
• 1:33:31
  And thanks for talking to us on Uncomfortable Conversation.
• 1:33:33
  Invite me back anytime.
• 1:33:34
  This was a lot of fun.