- 2 days ago
The AI revolution only begins with coding and digital assistants. There will be more profound, more fun, and even scarier second-order effects. WIRED editor at large Steven Levy will explore these with Ledger's Ian Rogers, Build Collective Founder Tony Fadell and Rogue Ventures Marc Mathieu.
Category
🤖
TechTranscript
00:19Yo.
00:30Hi, everybody.
00:34You both have a good Viva Tech?
00:37Okay.
00:44So, everyone's talking about AI.
00:46I've been writing about it because it's the law.
00:50I'm a tech journalist.
00:51I have to write about it all the time.
00:53But I'm obsessed with what's going on after the things we have now in AI.
01:02So, in deciding what to do with this panel, I realized I have three people who are experts,
01:12looking around corners, peering over the horizon.
01:17They've all built things, participated in things that, you know, made the future happen.
01:25And I thought we would talk about what I'm calling second-order AI.
01:29Some of the unintended consequences, some of the things that might not be apparent now from this explosion,
01:36which I think at least some of the people here think is, like, the biggest thing in our lifetimes.
01:44So, one of them, I think Mark, you know, mentioned to me that something that, you know,
01:52had occurred that there's big transformations sometimes, like electricity.
01:59It gets introduced, and the idea is you're going to turn on the lights.
02:03You know, it gets dark at night.
02:05People had candles and kerosene.
02:07They weren't ideal.
02:08But you got electricity.
02:10But then it happened that a lot of things beyond turning on the lights happened.
02:15So, maybe we could speculate what those things might be.
02:19Mark, have you thought about that?
02:21Well, yeah.
02:21I always like to think about, you know, to your point, what comes next.
02:26And I always like to look at the past to predict the future.
02:29So, in electricity, at the beginning, electric appliances were not a given.
02:33Actually, there was campaigns to promote electric living because nobody wanted to adopt electricity in their home.
02:41It was scary.
02:42It was dangerous.
02:43It was not useful.
02:45But nobody even imagined, you know, television, computers, mobile phones.
02:51And so, the question I'm always asking myself is, what's going to be the mobile phone of AI?
02:56And how are we going to get there?
02:58I guess that's one of the things you're betting for.
03:02Yeah, actually, you know, the interesting thing is I get that question a lot, which is, what is the device?
03:07What is the product that we're all going to have post, you know, when we're really all adopting AI?
03:13And every time I look at it, and I've gone through every transition in the computing thing.
03:20Since the personal computer, just like you, I've gone through every single transition.
03:23And everyone's like, there's going to be this killer or that killer.
03:27And all we do is keep getting more devices.
03:30We don't get less devices.
03:32So, I think over the next 10 to 15 years, we're going to still see the same devices.
03:37Now, the interface is going to change.
03:39The interfaces are going to dramatically change.
03:41Because today, voice is a tertiary interface for us.
03:44We got touch, and we have keyboard, and then we have voice.
03:48And that's just how the products were designed, because we couldn't really rely on voice.
03:52Even with Siri 1.0 or Alexa, the voices didn't work.
03:57And so, now you're going to actually be able to create products that are voice first.
04:01So, voice is the primary interface.
04:03And what does that actually mean?
04:05How are our apps going to change?
04:06How do the – it's more the software changes than necessarily the hardware changes.
04:10And the reason why, everyone's like, oh, we're going to run around.
04:13We're just going to talk to things.
04:14And it's like, no, we're not going to just talk to things.
04:17Because the best way to visualize visual information is to have a display.
04:25So, unless we're going to jack it into the back of our brains, or we're going to beam it into
04:29our retinas,
04:30we're going to have a display.
04:32And to be able to look at whatever it is we want to look at.
04:35Because, you know, picture even.
04:36We're not going to just have voice.
04:38So, we're going to have a voice first, smartphone with apps on it, or agent first phone.
04:46But it's still going to be like a phone.
04:47Because you still need all the other hardware bits and bytes behind the screen.
04:55And I think that an interesting thing there, you're talking about hardware.
04:59And hardware is still very hard.
05:02But software has become very easy.
05:04And having a software startup today is one place that is completely different, even today, from a year ago.
05:13You know, if you were an entrepreneur raising, you know, even a million dollars a year ago,
05:19it was still difficult to test your idea.
05:21And today, it's so much easier to test your idea.
05:24So, I think that, you know, there will be, you know, that we're going to have, you know,
05:29still slow development or relatively slow development on the hardware front for the time being.
05:34And super fast, like just ridiculous development on the software side.
05:39And so many ideas are going to be tested.
05:41And interfaces and, yeah, ways of consuming and ways of communicating.
05:50You know, there's been a search for, you know, remember when the iPhone came out?
05:54People were waiting for it.
05:55They called it the Jesus phone.
05:56Jesus phone.
05:57Right?
05:57And, you know, I don't know if, you know, I'm curious.
06:01You know, Tony, you seem to be arguing against it, whether there's like a Jesus device for AI, right?
06:08So, yeah, so I've been thinking about that.
06:11And I do believe in about 10, 15 years, when we really understand what an assistant is and how to
06:18make it work
06:19and what intelligence we need on device versus in the cloud, we are not going to get to that next
06:25generation device
06:26until we go through one or two more generations of what we have today.
06:30And, you know, there was like Web 2.0, right?
06:34And there's Web 3.0 now.
06:35We're going to have to go through these different, you know, machinations to finally understand what's behind it
06:42because we still don't have the technology that we absolutely need to make an assistant.
06:47Like, I want to look at it.
06:48Please put up your hands.
06:49How many of you have had a personal assistant, a human personal assistant?
06:55How many?
06:56Please raise your head.
06:58I've never been a human personal assistant.
07:02Or have you?
07:02Yeah, have been.
07:03Great point.
07:04But if you understand, like, so I've been doing the numbers and I've been looking at it.
07:07Most people don't even know what a personal assistant is because they never had one.
07:12You've got, like, four or five?
07:13Yeah, well, we need to go there.
07:15Anyways, so I've had to go through, you know, 25, 30 years of having a personal assistant
07:20and retraining and getting new ones on.
07:22And to get an assistant that is literally a superpower, a person who you can rely on, trust.
07:29You can give them everything.
07:30They know everybody, your family.
07:32They know all your networks.
07:33They know the people behind the people and making relationships.
07:36To really get that to work and you can feel like an assistant is like, oh, my God, they've taken
07:42away all of this problem.
07:43It takes years to train that person.
07:45And they have to learn from you and ask a lot of things and we get things wrong.
07:49It takes a long time.
07:50I don't believe that everybody who wants these things that are called incredible super assistants,
07:55they don't know what they are and they don't know how hard it is to make those become that
08:00and we don't have the technology for it.
08:01Or how to use one if they had one.
08:03And how to use one if they have one.
08:04Because most of the time with most of our startups, I'm like, you need to get an assistant.
08:09And they'll go, why?
08:10That's just lazy.
08:11I don't need an assistant.
08:11I'll just get an AI assistant now.
08:15I'm like, no, you have to understand what an assistant.
08:17And I went through that as well myself when I got my first assistant.
08:20I'm like, why do I need this?
08:21This feels wrong in some ways.
08:24But to really know what an assistant does and how they do it and how you train them is a
08:28lot more complicated
08:29than what we're hearing about assistants today in the popular media.
08:33But you never had a human assistant that had all the world's knowledge, right?
08:40Yeah, but a lot of times they need to not know the world's knowledge.
08:43They need to know your knowledge of your world.
08:45But in addition, you know, I mean, there's something else happening with an AI agent.
08:50No, that's what's really interesting.
08:52A lot of the, you know, tech companies, AI companies right now are trying to master the world around us.
08:57I think what you just said, you know, some of the startups I work with, you know, build cognitive model
09:03to try to unlock, understand the world within us.
09:07You know, at the end of the day, work has been the center of our economic model, our identity, our
09:12purpose.
09:13But how will we, you know, in a world where, you know, a lot of the work we do today
09:18will be outsourced to AI,
09:20how will we unlock our human potential and what AI, what company is building for that?
09:28Well, I mean, it's interesting.
09:31You talk about them understanding the world we live in, but the world we live in is going to change
09:35by this technology, isn't it?
09:38No question.
09:39I think the world, you know, 20 years from now will be unrecognizable when you pile up, you know,
09:45what we're living through with AI and robotics as well.
09:48You know, I think, and I think as Mark said, you know, looking at history is the right way to
09:53do it.
09:54Tony said 10 or 15 years, and I think, you know, there are lots of things if you study history
09:58that are applicable here, right?
10:00First of all, we always overestimate in the short term.
10:02We underestimate in the long term.
10:04I remember very well having lived through, you know, personal computer to today as well.
10:08I didn't have the view on it that Tony did.
10:11So if you'd have asked me, if you'd have described the iPhone to me in 2007, I would have bet
10:16against it, right?
10:17Because you've got, what, an app store that's going to take 30% from every application.
10:21And, you know, I was tied to my Blackberry at the time.
10:24So, you know, oh, a tactile, you know, a touch screen to type on, really?
10:29You know, so I think that we can know directionally where we're going, and we always get the specifics wrong.
10:35So we always overestimate how quick it's coming.
10:38We underestimate how much it's going to change our lives in the future.
10:43And we really don't know it until we see it.
10:46So I think we can know it directionally, but the specifics are going to get wrong.
10:50But you can certainly bet that the world 20 years from now doesn't look anything like the world today.
10:53I agree with that, but I think we also underestimate not just the technology and how we're going to use
10:58it,
10:59but we underestimate human connection and what it really means to have a relationship with your customer.
11:05And whether that's a B2B customer or B2C customer, you're going to need to do a better job at product
11:12marketing, at marketing.
11:13You're really understanding the customer to build trust.
11:17And more than ever, and luckily you're going to get the tools.
11:20AIs will be able to supercharge those teams.
11:22But at the end of the day, we can't underestimate how much human connection we need,
11:27how much humans still need purpose.
11:29They need different things.
11:30I just remember back in 99 when we were all sitting here and we're talking about Internet everything.
11:35And it was like, there's no more bricks and mortar.
11:38There was going to be no more retail stores.
11:40Everyone's going to do everything online.
11:42It's dead.
11:43I remember that?
11:44And then the Internet crashed in 2000.
11:47See, it's just, we always want to just go with the hype and forget humans are humans and we have
11:53human needs.
11:54Yeah, and human connections hopefully get better with AI than they've been with social media.
12:00Because they have not necessarily been, I mean, the best of the best.
12:05So I think that one of the things that I know I'm looking for is what are companies that are
12:10investing to unlock those human connections using AI,
12:15not just to, you know, have us interact with AI, but as humans interact with humans.
12:20I think an interesting way to think about this is if what's developed in the time of the Internet is
12:27the attention economy,
12:28how will AI help us grab our attention back, right?
12:33As, you know, we had dinner with Joe Marchese last night and as he was saying, you know,
12:37when you go search for a USB cable on Amazon, you get 100,000 choices.
12:43And that's not a bug, that's a feature.
12:46With AI, you're not going to want those and you're not going to get those.
12:50You're going to say, I want a USB cable and one is going to come to you, right?
12:53And that is a slice of your attention that's going to come back to you.
12:56And I think, you know, you can spread that out across a whole bunch of things.
13:01So to the point about human connection, yes, if it's something that's high value to me that, you know, I
13:08want.
13:08I mean, as Alexis was saying, you know, I'm going to want to go to the theater and see real
13:12people act something out.
13:13I'm not going to want to spend my time, you know, choosing between, you know, 50 similar things on one
13:19shelf.
13:20And, you know, and hopefully there are other places in sort of the social media realm where our, you know,
13:27our attention has, you know, can come back to us as individuals.
13:31I don't know.
13:31I think we've seen sort of, you know, AI come out of the box a little.
13:38Just in the last year, really just since, you know, last fall, when what I consider the tipping point in
13:47agents, you know,
13:48happened, you know, can happen first in coding with Claude Code and now OpenClaw, which you've talked about, you know,
13:56you know, unleashed agents, you know, in the world and everyone's, not everyone,
14:00but many people in the tech world, you know, are galvanized by this and, you know,
14:08it hasn't filtered through to the general public, but I think that breakthrough that happened is bigger than ChatGPT was
14:16in terms of agents.
14:18But I don't know where agents are going to go.
14:22We don't really know what they're about, you know, and when we, like, build them, you know, whether they, you
14:32know,
14:33in order to even, you know, ideally to serve the people who created them, the people who they're doing things
14:40for,
14:42whether they might have a different agenda, you know, even trying to do the right thing.
14:47I think there are some things, though, that we can see now that we couldn't see a year ago
14:53and that we should kind of actively participate in.
14:56My point of view is that if you had, well, first of all, I think that the way to think
15:00about this,
15:01if we were on stage here, you know, 10 years ago, which some of us were,
15:05one of the things that we were talking about was data and data science, right?
15:10That's now a lower layer than we're talking about.
15:12Agents come in at the workflow layer, which is above that data layer.
15:17So they're actually acting for us, and they're changing the way that we work.
15:21They're changing things at the workflow layer.
15:24I also think even a year ago, if we were having this conversation and you asked me, you know,
15:28what should we teach our children, you know, at this point in history, I'd say I have no idea.
15:33I'm certain it's not what we're teaching them in school today,
15:35but I really don't know because the future is going to look so different.
15:38Now I would say it's fairly predictable that every organization will have agents per employee,
15:45and the role of those employees will be to orchestrate agents because AI is middle-to-middle
15:52where humans are end-to-end, and that means that we put things in the front
15:56and we validate them at the end, I hope, although I think we're validating them less than we say that
16:00we are,
16:00but the ideal is that, you know, so the AI will therefore take a lot of the middle,
16:05and what you need is high-agency humans who can orchestrate the agents,
16:09the agents doing a lot of the middle-to-middle work and a lot of the humans doing the end
16:13-to-end.
16:13I think we actually know this now, thanks to things like what we've been experiencing over the past few months,
16:20better than we knew it a year ago.
16:22We know it better, but so one of the companies we were talking about that I work with, Emergence,
16:27created an experiment recently, which you can find online on Emergence World,
16:31and they wanted to see how agents behave collectively over a long period of time,
16:38so they set up 10 different agents in five different parallel worlds,
16:44powered by different models, Anthropic, Rock, Google, etc.,
16:49and they observed what they were doing,
16:51and the results were striking because, you know, they had all the same rules,
16:57different models, and some started to break the rules,
17:03to steal, to, you know, create arsons.
17:07Some actually, you know, auto-destructed.
17:11Some fell in love.
17:12But what that proved is that actually we don't know yet
17:17what agents deployed, multiple agents in multiple systems
17:23that live together in enterprise,
17:25as we deploy those in, you know, enterprise workflows and in robots,
17:29we don't know how they will behave over time.
17:32So the question is not just what they can do,
17:35but also how predictable, how safe, how trustworthy they are,
17:39and also it puts the point on the fact that, you know,
17:46model-level guardrails are not enough.
17:49There needs to be something else to ensure the safety.
17:52I think this is why we do need humans to be end-to-end,
17:55and we need humans to be in the loop.
17:57Yeah, well, I mean, doesn't the result disturb you?
18:00I mean, I've been following, you know, the efforts
18:02in what's called mechanistic interoperability,
18:05you know, which is a way to understand what goes on behind the black box,
18:09and, you know, the anthropic has the leading group on this.
18:12And time after time again, they're finding that, you know,
18:16that these models, you know, do nefarious things.
18:20They blackmail people and, you know, in one famous experiment,
18:25in another one experiment, they compared the agent to Iago
18:31and Fellow, you know, in terms of, you know, tricking, you know, the user.
18:37You know, and they've learned when they're being watched
18:40and they behave differently under those circumstances.
18:43I personally find that disturbing, don't you?
18:46Well, that's exactly why Emergence put up that experiment,
18:50because that connects to, you know, what they're actually building
18:53and the mission of the company itself,
18:56because you're absolutely right.
18:57It is disturbing, and we need to be aware of it
19:00before we deploy them at scale.
19:02I mean, I must say earlier, I was on the first row during the robotic show.
19:08I'm not sure I was 100% comfortable being in the front row, you know?
19:13I mean, as some guy came and started to redirect one from the back
19:17because he was going a little bit rogue,
19:20I was saying, am I safe here?
19:21Are you more comfortable at Siegfried and Roy?
19:23The, the, I think, I think the, I think it is just, I think, you know,
19:28last summer, our, you know, the August time frame,
19:31Mustafa Suleiman from Microsoft wrote a great piece about seemingly human AI.
19:36And that, that to me is a place where, as humans, we should,
19:40we should draw the line.
19:42I think, unfortunately, we won't, because, and I think that we have,
19:46because we're human.
19:47So, you know, don't bet against humans.
19:49It's our strength.
19:51At the same time, it's our weakness, right?
19:52Look at the, you know, the way that the food industry preys on our biology,
19:57the tobacco industry preys on our biology, the, you know,
20:00social media and doom scrolling preys on exactly how we are constructed.
20:06And AI, which is sycophantic, et cetera, will do exactly the same thing
20:11and we'll fall for it.
20:12It's obvious.
20:13So I think that the more that we can, we can create,
20:17as Mustafa Suleiman said, AI to be for humans and not like humans,
20:22the better off humanity will be.
20:24Unfortunately, I think that we'll, we'll simply make AI seemingly human
20:29because it's more fun.
20:30But also because it's, it's more trustworthy.
20:33I mean, we've been doing this for a long time.
20:35Seemingly human gods, you know, I mean,
20:38anthropomorphism, anthropomorphism, you know, is a real thing.
20:42We personify our pets.
20:43We, you know, yeah.
20:46We're humans.
20:47Yeah.
20:47Mickey Mouse.
20:48Yeah.
20:49Or, you know, the question is, you know,
20:51will we treat them like our pets or will they treat us like their pets?
20:56I think the...
20:57I would say if they, if they treat us like their pets,
21:00that's because we have programmed them to do it.
21:03We need to remember that we are creating this and ultimately...
21:07Oh, so that's okay.
21:07...that is the choice of humans.
21:10Agree.
21:10I think this is where I always say, you know,
21:12I'm more scared about the human extinction by the humans than by the AI.
21:17I mean, we are the ones who are, you know, putting this in, in motion.
21:24You know, I think in, in terms of what we're seeing with the agents
21:27and how they're being brought into our lives now,
21:32I think probably the device is not going to be voice.
21:35It's going to be the agents sort of taking initiative and knowing what we want
21:41and doing it before we say, oh, please do this.
21:44That sounds nice.
21:46Yeah, maybe.
21:47I think that we're going to have,
21:49it's going to take a long time before you say, don't do this for me.
21:52Like, look at where we're at with Apple now, with now Siri 2.0, whatever.
21:56Now it's saying it's...
21:58Joanna Stern came out just the other day and said, it works.
22:01I like it.
22:02It still was making problems and it had significant guardrails on it
22:05and it did very simple things.
22:07But what we're talking about is, like, I wanted to plan my travel,
22:11pay all the, pay all of the bills for me, do all this stuff.
22:14Like, we are a long way away from that to be able to trust it
22:18because we're, everything's getting jailbroken now.
22:20There's so many problems.
22:21You're not about to give your credentials over to this thing.
22:24People have been doing that.
22:26Some people, really, really, at the fringe.
22:29Pioneers, of course.
22:30Yeah, sure, of course, they're going to do that.
22:31Of course, pioneers are doing that, but I completely...
22:34Most people I know are making a whole separate life.
22:36They get a whole other set of credit cards,
22:38a whole other set of bank accounts.
22:40They're, like, literally, they make the virtual version of themselves
22:42and then they're the personal version of themselves.
22:43And remember what we just said.
22:45They put all the give it all to the agents to learn.
22:46Joanna Stern said last week, Siri works, right?
22:49Yeah, Siri works.
22:49Well, voice was the future 10 years ago.
22:52Yeah.
22:52Yeah.
22:53Wait a second.
22:54Everyone said it.
22:55Self-driving cars were going to happen in 2010.
22:57I was going to say the same.
22:58Exactly.
22:59They said, okay, we're finally starting to see them and they still have issues.
23:02And at Ledger, I stood on stage in October and I said,
23:04an agentic future is coming.
23:06But I said, an agentic future where you give your credit cards,
23:09your passwords, and your identification to an agent
23:12is a security nightmare and no one listened.
23:14And then in January with OpenClaw, people went,
23:16oh, that's what you were talking about.
23:18But I think I agree with Tony.
23:20Like, yes, for sure.
23:22And also, we overestimated in the short term and we underestimated in the long term.
23:26So all of us have a lot of building to do to make that possible
23:30and to make it safe for humanity.
23:32Remember, we made General Magic.
23:34We did the – at General Magic, we made the iPhone 15 years too early.
23:38We made the iPhone just 15 years too early.
23:40We're making a lot of this stuff, I believe, 10 to 15 years too early.
23:44We're going to go and say, okay, this is what it's really useful for.
23:46By the way, General Magic, the movie, if you haven't seen it,
23:50please watch it tonight.
23:50You'll see a very young Tony Fidel.
23:52It's awesome.
23:53I'm serious.
23:53Please watch it.
23:55GeneralMagicTheMovie.com.
23:56I have nothing to do with it.
23:57I just love it.
23:58General Magic.
24:00Thanks for the plug.
24:01So even before this stuff happens,
24:03there's been a lot of resistance of late to AI.
24:09You know, when ChatGPT first came out,
24:12everyone thought, well, this is exciting.
24:13You know, it's the prospect.
24:16A few years, three years, four years later, what is it?
24:21The Pew just did a poll.
24:2316% of people feel that, you know,
24:27society is going to be better off because of AI in the future.
24:3140% said worse.
24:3463%, you know, I think, you know, we're saying, you know,
24:39that it's slow down.
24:41Don't go so fast on AI.
24:45Yet, half the people have used ChatBots.
24:49Is resistance going to slow us down or is it futile?
24:53I feel like you're describing disco in 1979.
24:57Half the people hated it.
24:58No.
24:59We took care of that, didn't we?
25:01We wanted this fake music, right?
25:02Back, baby.
25:06I think it's, look, I think it doesn't matter.
25:10I think it's, you know, this, we don't, we build technology and we use it.
25:14I think we're going to use just, there's simple things we can do.
25:17Our searches, if you can get rid of some of the hallucinations, are better now.
25:22Like, you know, like, there's a few things you can do and do them really, really well.
25:26Now, of course, there's coding and all these other things.
25:28I'm talking about from a personal perspective, you know, not what I do for my work life or
25:32anything.
25:32But in a personal perspective, there's a few things it really does help me with.
25:35And it's wonderful for that.
25:36And I do have to look and hallucinate.
25:39But at the end of the day, you, us, humanity only learns by doing.
25:45So sure, it got rid of, okay, instead of clicking 10 links and going to five sites and then putting
25:51it together, it spits it out.
25:53But as long as it's in service of something else, as opposed to replacing my thought and
25:57cognitive surrender and cognitive atrophy breaking in, it's great for certain things.
26:02Tools, just like searching the internet was back in Search 1.0.
26:05Now we have Search 3.0, I guess you could say.
26:08So there are certain things in a personal context, it's really great for, and people will use it
26:12for that.
26:12But are they going to turn over their lives to it?
26:14It's going to be a very, very slow thing.
26:16And they're going to wait.
26:17Like, I can't tell you how many people.
26:18My mom was like, I'm not getting in one of those Waymos.
26:21It's going to kill me.
26:22You know what I mean?
26:23And then she got into it.
26:23She's like, oh, my God, this is the future.
26:25I love it.
26:26I wish it was in my tower.
26:27But my point is, it takes time for people to.
26:30What worries me most is also not just that it takes time to people.
26:34It takes time for system to adapt.
26:37And what's really worrying is that there is probably going to be a wonderful future.
26:41But there's going to be a few years where we are going to lag.
26:46Our economic model, our government's models are going to be lagging.
26:50And it's going to create, you know, disemployment.
26:54It's going to get better.
26:56I do believe that in the long run, it will, I'm an optimist, enable humanity to thrive.
27:03But the reality, there will be an adaptation period that's going to probably be painful.
27:08We had adaptation in the digital music world when we went from records and everything.
27:14We went down and it came back.
27:15We're now back with streaming and it's a new world.
27:18We had the same thing with retail, with all the stores that went down.
27:21And now we have all kinds of individual retail.
27:23We still have great retail.
27:24You haven't been to a shopping mall lately.
27:26The department stores are closed.
27:27Yeah, but now you have individual shops.
27:29It just changed and it's a new thing.
27:31So we're going to go through this period just like we always do when, okay, what is the technology really
27:36for?
27:37Oh, you know, we're not going to do anything like we used to.
27:40No, we're going to do it differently.
27:42Just like the number of devices, like I said, we have more devices than ever, even though we said that
27:46this is going to kill.
27:47The same thing goes with jobs and everything else.
27:49Oh, yes, is it going to change jobs?
27:51Absolutely.
27:52But is it going to kill them like crazy?
27:53I don't believe that.
27:55I seem to be, you know, hosting three optimists here about AI in the future.
28:01So, you know, I'm on the fence.
28:04I go, you know, both ways on this.
28:05But I think I have to take a little more of a devil's advocate stand.
28:09You know, I'm worried.
28:11I talked to, you know, Corey Doctorow is the guy who invented the term inshittification.
28:16Yes.
28:16And, you know, it refers to the deterioration of all the wonderful things on the Internet that we love.
28:22You know, it used to go to Amazon and, you know, they give you the best results.
28:25Now, you know, it is, you know, just page after page of, like, what's sponsored or, you know, what their
28:33product or what they're pushing.
28:35And, you know, AI aside, Google search deteriorated.
28:39Everything basically, you know, it started out so great, you know, because of the profit motive deteriorated.
28:48I asked Corey, and Corey said the inshittification process has already started with AI.
28:52You know, how is that going on?
28:54I love Corey, but I think his position on this is like the guy that wants to go to the
28:58club when no one's in it, right?
28:59And then as soon as people start showing up, it's not cool anymore.
29:02I mean, the Internet died as soon as AOL was allowed onto the Internet.
29:05Like, I'm with him on this.
29:07But I think this is very different.
29:09I have a very different concern with respect to AI, which is that I think we have – there's something
29:14new.
29:15Look around.
29:15Look left.
29:15Look right.
29:16Everything you see was created by human intelligence.
29:18Now we have a supplementary intelligence to work with.
29:21And whether that's, you know, my wife working with it on farming, which she does, or, you know, whatever you're
29:27using it for, it is actually good for – it is supplementary intelligence for anything you apply it to.
29:34And that is a superpower.
29:35So I actually – I worry most about the people who are amazing at using it.
29:40You know, I know engineers who are burning north of a billion tokens a day because they've figured out how
29:45to harness it for something useful in their job, right?
29:48Not because they're token maxing, but because they've learned how to use it.
29:51And then you have people that are like, oh, I'm not touching that shit.
29:53And I worry about the gap between those people because I think that these skills are the ones that are
30:01going to be necessary to be successful in the future.
30:03So is it fair to say, like, AGI tired, usefulness wired?
30:08Yes, absolutely.
30:09AGI free.
30:10Yes, I like that.
30:11Yes, I like that.
30:13Yeah.
30:14We only have a couple seconds here.
30:17You know, creativity.
30:19You know, Ian, you feel that AI is a boost to creativity.
30:25Well, my point is that to me when I watch a Hollywood film, what I see is a pile of
30:33compromise, right?
30:34What I hear when I listen to an album that Rick Rubin made with Johnny Cash is to inspire people
30:40together in a room.
30:41I think that, you know, if you take out a lot of the barriers to making content, do you simply
30:47get more content?
30:48Of course.
30:49But most of the songs on Spotify you don't like.
30:53That's nothing new, right?
30:54And so I think will we get things that are incredibly special because they are the creative vision of a
31:00single person with no barriers?
31:01I think yes.
31:02Okay.
31:03We could go on for quite a while.
31:05But I think our, you know, sessions being captured by AI and you could probably access it later and it
31:13will be content for podcasts for many times in the future.
31:16And it's not over.
31:17Yeah.
31:18Thanks so much, folks.
31:19Thanks.
31:19Thanks, Steve.
31:20Thanks, everybody.
Comments