Latest / General Magic & The Soul of Tech / Ep. 3 | What Orange’s Chief AI Officer Learned from General Magic | Steve Jarrett

Transcript

Ep. 3 | What Orange’s Chief AI Officer Learned from General Magic | Steve Jarrett · General Magic & The Soul of Tech

• 0:01
  SPEAKER_03: This is Mike Stern, and this is Tom Hersh-Hirshenson.
• 0:05
  SPEAKER_03: Welcome to General Magic & The Soul of Tech.
• 0:10
  SPEAKER_00: We believed we were building something that would change the world, and in some ways we did.
• 0:15
  SPEAKER_00: In others, we didn't understand what we were setting in motion.
• 0:18
  SPEAKER_00: The story of General Magic isn't over.
• 0:21
  SPEAKER_00: We're still living inside what it started.
• 0:27
  SPEAKER_03: Our guest today is Steve Jarrett,
• 0:29
  SPEAKER_03: a guy who's at the heart of deployment of AI,
• 0:32
  SPEAKER_03: both in the enterprise and to the public.
• 0:34
  SPEAKER_02: And if you like having a web browser on your phone or using a mobile app store,
• 0:40
  SPEAKER_02: or if you just happen to enjoy music on your mobile device,
• 0:43
  SPEAKER_02: you have a debt of gratitude to Steve Jarrett,
• 0:45
  SPEAKER_02: who played an integral part in all three of those things being in the universe
• 0:50
  SPEAKER_02: of mobile functionality.
• 0:52
  SPEAKER_03: Yeah, welcome, Steve.
• 0:53
  SPEAKER_03: You're the head of AI at Orange,
• 0:54
  SPEAKER_03: and that's an incredible catbird seat on looking at what it means to deploy AI both
• 1:00
  SPEAKER_03: in the enterprise and in a huge enterprise that faces,
• 1:03
  SPEAKER_03: what, tens or hundreds of millions of users.
• 1:05
  SPEAKER_02: Yeah, 300 million.
• 1:07
  SPEAKER_03: Yeah.
• 1:08
  SPEAKER_03: So I'd like to kick off by asking,
• 1:10
  SPEAKER_03: what can you tell us about AI governance,
• 1:13
  SPEAKER_03: what you're doing to safeguard the way it rolls out both inside your company and
• 1:18
  SPEAKER_03: to your customers?
• 1:19
  SPEAKER_01: Sure.
• 1:19
  SPEAKER_01: And it's great to see you guys.
• 1:20
  SPEAKER_01: And thanks for having me on.
• 1:22
  SPEAKER_01: Orange, 300 million customers across 26 countries,
• 1:25
  SPEAKER_01: and the range is France to Democratic Republic of Congo.
• 1:30
  SPEAKER_01: So incredible scope of people's ability to pay,
• 1:34
  SPEAKER_01: our ability to understand which language they're speaking.
• 1:37
  SPEAKER_01: Infrastructure availability.
• 1:39
  SPEAKER_01: And so the challenge is really extreme in trying to provide technology that works
• 1:43
  SPEAKER_01: for people who are as remote as those in rural DRC.
• 1:48
  SPEAKER_01: The way we do it in the company, the majority of the value is actually initially for employees.
• 1:53
  SPEAKER_01: So making them more efficient, making it easier for them to do their daily jobs.
• 1:56
  SPEAKER_01: We spend a lot of time training employees in what we call responsible AI.
• 2:01
  SPEAKER_01: And for us,
• 2:02
  SPEAKER_01: that's not just being ethical,
• 2:03
  SPEAKER_01: but it's also really having people understand, don't use AI if you don't have it.
• 2:08
  SPEAKER_01: You should try first with a simpler method because it's more explainable,
• 2:11
  SPEAKER_01: it's more repeatable, it's more deterministic.
• 2:13
  SPEAKER_01: It also costs the company a lot less money and it's going to generate a lot less CO2.
• 2:17
  SPEAKER_01: And that's a hard lesson to learn because the tools are so powerful and easy to use
• 2:21
  SPEAKER_01: that people get lazy feedback.
• 2:23
  SPEAKER_01: And they often just start defaulting everything to the most powerful models.
• 2:27
  SPEAKER_01: We have to learn that lesson in Africa because people can't afford to use those models.
• 2:32
  SPEAKER_01: And we face even more extreme challenges like African rural languages spoken
• 2:37
  SPEAKER_01: in these And then we have all of those plus the affordability
• 2:40
  SPEAKER_01: and accessibility challenges in all the African countries.
• 2:42
  SPEAKER_01: Wondering with respect to the...
• 2:52
  SPEAKER_02: I mean,
• 2:54
  SPEAKER_02: you highlighted the notion of people getting lazy and not necessarily engaging
• 2:58
  SPEAKER_02: the simplest tool first.
• 3:00
  SPEAKER_02: I'm wondering the extent to which you look at anticipating social effects that have
• 3:05
  SPEAKER_02: an ethical component.
• 3:07
  SPEAKER_02: How does that come into your thinking personally and what you've seen at Orange?
• 3:11
  SPEAKER_01: Yeah, it's a great question.
• 3:12
  SPEAKER_01: And I should preface this by saying these are my personal opinions,
• 3:15
  SPEAKER_01: so not always the opinion of the company.
• 3:18
  SPEAKER_01: In France, people are employed for as long as they want to be employed by the company.
• 3:22
  SPEAKER_01: And in the U.S., we see these very, very large layoffs occurring, oftentimes AI.
• 3:26
  SPEAKER_01: Is used as the excuse for a large layoff.
• 3:28
  SPEAKER_01: But certainly the challenge is real of the people that are happy
• 3:32
  SPEAKER_01: to adopt these tools and are much more efficient and effective in their daily jobs.
• 3:36
  SPEAKER_01: Maybe that's 20% of the employees,
• 3:38
  SPEAKER_01: whereas the rest are either fearful or they're not comfortable learning new tools.
• 3:43
  SPEAKER_01: And so there's a lot of work necessary to make them feel comfortable.
• 3:46
  SPEAKER_01: And the way we describe that is these are tools to help automate the parts of
• 3:51
  SPEAKER_01: the job that you find really boring or that aren't fun or interesting.
• 3:55
  SPEAKER_01: You want to have people get more generally more comfortable with using the tools.
• 3:58
  SPEAKER_01: And also the underlying tools are evolving so rapidly.
• 4:02
  SPEAKER_01: It's fastest moving technology I've ever seen.
• 4:03
  SPEAKER_01: It's very hard to predict beyond 12 months or even six months what's going to happen.
• 4:08
  SPEAKER_03: Does government have a role to play?
• 4:09
  SPEAKER_03: How does EU regulation,
• 4:11
  SPEAKER_03: which we don't have in the U.S.,
• 4:12
  SPEAKER_03: how does that impact how you think about it and how you deploy?
• 4:16
  SPEAKER_01: Yeah, that's also a really hard balance.
• 4:18
  SPEAKER_01: I mean, generally, Europe is overregulated.
• 4:21
  SPEAKER_01: There was a recent report by Draghi.
• 4:23
  SPEAKER_01: He's hired by the European Commission to write a competitiveness report for Europe.
• 4:27
  SPEAKER_01: And the big complaint was, hey,
• 4:29
  SPEAKER_01: you're massively overregulating and I'll So the problem with the AI Act,
• 4:35
  SPEAKER_01: for example,
• 4:35
  SPEAKER_01: there were some good things in it,
• 4:37
  SPEAKER_01: like identifying use cases that are classified as high risk, where,
• 4:40
  SPEAKER_01: for example,
• 4:41
  SPEAKER_01: you don't want to use AI models to do entirely independently doing credit cards.
• 4:45
  SPEAKER_01: So that part of the AI Act made sense.
• 4:48
  SPEAKER_01: Another thing that makes a lot of sense in the AI Act is that they were requesting
• 5:01
  SPEAKER_01: That the model providers give information about how much CO2 is generated in
• 5:05
  SPEAKER_01: the consumption and the creation of the model and also how much gets generated
• 5:09
  SPEAKER_01: when you run the model.
• 5:10
  SPEAKER_01: And so that's a really nice natural way for people to learn
• 5:14
  SPEAKER_01: that I'm actually impacting the world.
• 5:16
  SPEAKER_01: By choosing a much more expensive, more powerful model when I don't have to.
• 5:19
  SPEAKER_01: But then there were a bunch of things that really discouraged innovation.
• 5:23
  SPEAKER_01: Like if you customize a model,
• 5:25
  SPEAKER_01: which is what you do in the open source community,
• 5:27
  SPEAKER_01: you take these open weight models that are really popular.
• 5:30
  SPEAKER_01: And in fact, the challenge.
• 5:31
  SPEAKER_01: And so that just put a massive damper on the open
• 5:46
  SPEAKER_01: The AI model community in Europe.
• 6:03
  SPEAKER_03: Trump has, you know, giving credibility to anything Trump says is risky.
• 6:07
  SPEAKER_03: He spotted the idea that the U.S.
• 6:08
  SPEAKER_03: government would be a gatekeeper and vet models before their release.
• 6:12
  SPEAKER_03: Is that feasible given the impedance mismatch between the speed of
• 6:16
  SPEAKER_03: a government committee and the way the private sphere innovates?
• 6:20
  SPEAKER_03: Is that workable?
• 6:21
  SPEAKER_03: Are there other approaches that would make more sense?
• 6:23
  SPEAKER_03: Because Tom and I have bandied about the idea of a public health approach to
• 6:27
  SPEAKER_03: the regulation of AI.
• 6:28
  SPEAKER_03: In the same way we have an FDA that deals with food safety or drug safety.
• 6:31
  SPEAKER_03: Yeah.
• 6:33
  SPEAKER_03: Is that an appropriate model?
• 6:34
  SPEAKER_03: But is it something government is capable of?
• 6:37
  SPEAKER_03: How do you help make it capable?
• 6:38
  SPEAKER_01: Yeah.
• 6:39
  SPEAKER_01: Yeah.
• 6:39
  SPEAKER_01: I mean, generally,
• 6:40
  SPEAKER_01: I think the retrospective fines for misbehavior is much easier to enforce
• 6:46
  SPEAKER_01: than trying to be super clever in advance about all defending against all
• 6:50
  SPEAKER_01: the possible bad outcomes.
• 6:52
  SPEAKER_01: The latter is what Europe tried.
• 6:54
  SPEAKER_01: With the AI Act, they tried to anticipate all the potential bad outcomes and regulate those.
• 6:58
  SPEAKER_01: But the other question that I think you're asking is one of should the U.S.
• 7:03
  SPEAKER_01: encourage closed commercial-only models versus open-weight models?
• 7:08
  SPEAKER_01: And that is a topic I flip-flopped on a few times.
• 7:11
  SPEAKER_01: At first,
• 7:12
  SPEAKER_01: my sense was that you needed to protect against the risk of these very,
• 7:17
  SPEAKER_01: very powerful models falling into the wrong hands.
• 7:19
  SPEAKER_01: That was my opening position.
• 7:21
  SPEAKER_01: Then it evolved into the position of, look, it's going to happen.
• 7:23
  SPEAKER_01: No matter what we do.
• 7:25
  SPEAKER_01: And so better to have all of the models in the open so that like in
• 7:29
  SPEAKER_01: the cybersecurity domain,
• 7:31
  SPEAKER_01: people in the open source community,
• 7:32
  SPEAKER_01: all the white hats can work together to very rapidly identify problems
• 7:36
  SPEAKER_01: and solve them and fix them.
• 7:37
  SPEAKER_01: But I've changed my point of view yet again because
• 7:41
  SPEAKER_01: the commercial models are getting very, very powerful now.
• 7:44
  SPEAKER_01: So the model from Anthropic called Mythos that was initially very focused
• 7:49
  SPEAKER_01: on cybersecurity,
• 7:50
  SPEAKER_01: it was able to find And so,
• 7:53
  SPEAKER_01: that kind of power, I think you really don't want to provide to organized crime.
• 8:08
  SPEAKER_01: And especially to nation states that aren't friendly to the West.
• 8:12
  SPEAKER_01: So I've gone back to my position on commercial clothes models.
• 8:16
  SPEAKER_01: And also my position has been held by the fact that
• 8:18
  SPEAKER_01: the open weight model community is really fast moving and doing a great job.
• 8:23
  SPEAKER_01: And so the gap between functionality is between six to 12 months.
• 8:27
  SPEAKER_01: And I think that's a reasonable gap for the benefit of having
• 8:30
  SPEAKER_01: the more secure versions that are closed in the right hands.
• 8:33
  SPEAKER_01: And there will be architectures that will emerge that will get us closer
• 8:37
  SPEAKER_01: to what we have in our brains.
• 8:39
  SPEAKER_01: And when that occurs, it's an even bigger transformation.
• 8:44
  SPEAKER_01: I mean,
• 8:44
  SPEAKER_01: the notion that the power of the best models today that you have running on a cloud,
• 8:49
  SPEAKER_01: if that runs in your phone, And in everybody's phone, that's a game changer.
• 8:55
  SPEAKER_03: If you think about the impact of having social media on your phone,
• 8:58
  SPEAKER_03: having internet access on your phone,
• 9:00
  SPEAKER_03: having an inference model on your phone seems to me orders of magnitude more risky.
• 9:04
  SPEAKER_03: I mean, if there are 3 billion smartphone users in the world,
• 9:07
  SPEAKER_03: if all of them have super powerful AI on their phone,
• 9:10
  SPEAKER_03: that seems to me to open up far more risk than benefit, but...
• 9:14
  SPEAKER_01: You know,
• 9:15
  SPEAKER_01: one of the lessons from General Magic was you can envision this amazing world
• 9:18
  SPEAKER_01: where everyone's connected and that's got to be a universal good.
• 9:23
  SPEAKER_01: And there's all kinds of unintended consequences when that happens.
• 9:25
  SPEAKER_01: As we saw,
• 9:26
  SPEAKER_01: you mentioned the downside of social media,
• 9:28
  SPEAKER_01: because I was at Facebook for a while and early on and we thought that connecting
• 9:32
  SPEAKER_01: the world was a universal good and it proved to have a bunch
• 9:35
  SPEAKER_01: of unintended consequences.
• 9:37
  SPEAKER_01: For me, if you if you just think about the the change in
• 9:52
  SPEAKER_01: How education works,
• 9:53
  SPEAKER_01: where when we were kids,
• 9:54
  SPEAKER_01: all of us, you went to the library and the book wasn't there, you're out of luck.
• 9:57
  SPEAKER_01: And then the internet came and Google enabled us to search and find answers
• 10:00
  SPEAKER_01: to any question we had.
• 10:02
  SPEAKER_01: And now with AI models,
• 10:03
  SPEAKER_01: you have a teacher who can explain anything to you in many different ways
• 10:07
  SPEAKER_01: until you understand it and as simply as you need it to be or as visually
• 10:11
  SPEAKER_01: as you need it to be.
• 10:13
  SPEAKER_01: And that's extraordinarily powerful.
• 10:15
  SPEAKER_01: And so I think that is a universal good.
• 10:17
  SPEAKER_01: The idea that we can enable people in Africa to have this experience on their phone
• 10:22
  SPEAKER_01: to learn with the world's best teachers at their own pace
• 10:26
  SPEAKER_01: on whatever subject they're interested in,
• 10:27
  SPEAKER_01: that's just an enormously valuable thing for women to understand reproduction,
• 10:33
  SPEAKER_01: medical health, and kids to learn about science quickly.
• 10:37
  SPEAKER_01: And so I think that's very, very powerful.
• 10:41
  SPEAKER_01: It's also useful in that if indeed the model's in the phone,
• 10:44
  SPEAKER_01: then the data's not leaving the phone.
• 10:46
  SPEAKER_01: And so then you can have higher confidence
• 10:49
  SPEAKER_01: that all this really something stuff you may be talking to the model
• 10:52
  SPEAKER_01: about isn't going to leave the phone.
• 10:53
  SPEAKER_02: Yeah, it seems like a couple of things.
• 10:55
  SPEAKER_02: First of all,
• 10:56
  SPEAKER_02: with surveillance capitalism, our commercial activities are monitored and monetized.
• 11:02
  SPEAKER_02: But with AI, there's an even more intimate disclosure, right?
• 11:06
  SPEAKER_02: There is more robust information that people are conveying to AI than they are
• 11:12
  SPEAKER_02: with simply browsing websites.
• 11:14
  SPEAKER_02: So it seems like there's a risk there.
• 11:18
  SPEAKER_01: I think in search, the market collapsed very, very quickly to one provider.
• 11:23
  SPEAKER_01: And in social media, effectively the same thing happened.
• 11:26
  SPEAKER_01: What encourages me a bit more in the AI space is
• 11:30
  SPEAKER_01: that it appeared like OpenAI was going to be the front runner for a very long time.
• 11:35
  SPEAKER_01: But Google and Anthropic had caught up very rapidly.
• 11:37
  SPEAKER_01: And in the open source domain, the Chinese are the best.
• 11:41
  SPEAKER_01: Opposed to what happened in the search world and what happened in
• 11:44
  SPEAKER_01: the social media world,
• 11:45
  SPEAKER_01: it appears anyway that consumers will have a reasonable amount of choice for
• 11:49
  SPEAKER_01: the foreseeable future.
• 11:50
  SPEAKER_01: That's always a good outcome to avoid the kind of dangers that you're talking about
• 11:54
  SPEAKER_01: because then you can choose the provider who aligns best with your point of view
• 11:58
  SPEAKER_01: on privacy and security and so on.
• 12:00
  SPEAKER_01: I think if the market collapses to one, then we have a problem.
• 12:04
  SPEAKER_02: Yeah, duly noted.
• 12:05
  SPEAKER_02: I was going to say you're a rarefied company in terms of having not just seen,
• 12:09
  SPEAKER_02: but been intimately involved with inflection points and innovations
• 12:13
  SPEAKER_02: that have been tremendously impactful.
• 12:15
  SPEAKER_02: And I have two questions related to that.
• 12:18
  SPEAKER_02: The first is, what is that like from a kind of Yeah, that's nice of you to say.
• 12:24
  SPEAKER_02: Thanks, Tom.
• 12:51
  SPEAKER_01: The world is built by people.
• 12:53
  SPEAKER_01: Everything you see in the world has been imagined and built by somebody.
• 12:56
  SPEAKER_01: That was a big eye-opener in the very beginning of my career that you just have
• 13:00
  SPEAKER_01: to be bold enough to be willing to see the world in a different way and
• 13:05
  SPEAKER_01: to see what is missing and then go build it.
• 13:07
  SPEAKER_01: And that was a really radical thought for me.
• 13:10
  SPEAKER_01: I grew up totally middle class, playing sports and video games.
• 13:14
  SPEAKER_01: And the notion that the world is built and created by people just like yourself was
• 13:21
  SPEAKER_01: a big, big
• 13:22
  SPEAKER_00: mental moment for me.
• 13:23
  SPEAKER_01: So the experience is generally the really good ideas are the ones that as soon
• 13:30
  SPEAKER_01: as you hear it, you think this has got to exist.
• 13:33
  SPEAKER_01: I can't be the only one who's thought of this.
• 13:35
  SPEAKER_01: This must exist somewhere.
• 13:36
  SPEAKER_01: I just haven't found it.
• 13:38
  SPEAKER_01: I think...
• 13:38
  SPEAKER_01: For me, what's most fun is the first time that it actually works is really a sense
• 13:44
  SPEAKER_01: of wonder whenever you see this thing that you've thought up with a bunch
• 13:48
  SPEAKER_01: of your friends and that you've had a bunch of sleepless nights and a bunch
• 13:52
  SPEAKER_01: of failures at making it work and then it finally does work the first time.
• 13:55
  SPEAKER_01: That's a really amazing feeling that's hard to beat.
• 13:58
  SPEAKER_01: The other really great feeling that's hard to beat is if you're lucky enough to work
• 14:01
  SPEAKER_01: on something that people use in their daily lives and without knowing
• 14:06
  SPEAKER_01: that you've worked on it,
• 14:07
  SPEAKER_01: someone tells you how great something is, is also just an amazing moment.
• 14:13
  SPEAKER_02: Sting tells a story about he was in the hospital and I believe it was his You used
• 14:18
  SPEAKER_02: a word that struck me
• 14:39
  SPEAKER_02: What you were describing where you talked about something that you worked on with
• 14:42
  SPEAKER_02: and you didn't say colleagues or coworkers, you said friends.
• 14:46
  SPEAKER_02: And I wonder if that's been a through line in your career in innovation,
• 14:50
  SPEAKER_02: the relationships you've had with the people that you've tended to work with friends
• 14:55
  SPEAKER_02: as opposed to people who were simply or only colleagues or coworkers.
• 14:59
  SPEAKER_01: Yeah, that's correct.
• 15:00
  SPEAKER_01: When you're trying to build things that have never existed before,
• 15:03
  SPEAKER_01: the amount of commitment is so extreme that even if you didn't start out as friends,
• 15:08
  SPEAKER_01: you become friends through the process because you're just together
• 15:12
  SPEAKER_01: in these incredibly intense situations where things are regularly failing.
• 15:17
  SPEAKER_01: And so you're regularly confronted together with your weaknesses.
• 15:22
  SPEAKER_01: And so the outcome of that is always or hopefully often, very often friendships.
• 15:27
  SPEAKER_01: And And so you were asking before about what my advice might be to people who want to innovate.
• 15:32
  SPEAKER_01: I think if I look back at my career,
• 15:34
  SPEAKER_01: when I was working with the most diverse and interesting and curious people,
• 15:40
  SPEAKER_01: that's when we had the best outcomes.
• 15:42
  SPEAKER_01: Because if you have a super diverse And so having a really diverse team is critical
• 15:45
  SPEAKER_01: so that you're more likely to get early signal from everyone's different perspective
• 15:49
  SPEAKER_01: on what's going to work and what isn't going to work.
• 15:52
  SPEAKER_01: And also having really
• 16:09
  SPEAKER_01: And curious, smart, hardworking people also leads to just a much more fun workplace.
• 16:14
  SPEAKER_01: I mean, when we all worked together at General Magic,
• 16:16
  SPEAKER_01: there were all kinds of hijinks that were happening
• 16:17
  SPEAKER_01: because people were really quirky and fun.
• 16:21
  SPEAKER_01: And so that moves from being a job to being something fun and rewarding at the same time.
• 16:29
  SPEAKER_03: You guys were the key pranksters at General Magic.
• 16:32
  SPEAKER_02: I can neither confirm nor deny that we crashed.
• 16:35
  SPEAKER_02: What were your favorite goofs?
• 16:36
  SPEAKER_03: Because I wanted to circle back a little bit.
• 16:38
  SPEAKER_03: You helped to invent with Tony at Phillips, the first App Store.
• 16:41
  SPEAKER_03: Now it's a trillion-dollar business.
• 16:42
  SPEAKER_03: I mean, that must blow your mind.
• 16:44
  SPEAKER_03: What were the really mind-boggling goofs you did at Magic?
• 16:48
  SPEAKER_03: But how did it feel to have something that started off in that little acorn become this oak?
• 16:52
  SPEAKER_01: Yeah, well, I mean,
• 16:53
  SPEAKER_01: that's a good example of we convinced developers at General Magic to write apps
• 16:58
  SPEAKER_01: for these devices we were creating.
• 16:59
  SPEAKER_01: And the app developers were really struggling to make any money because
• 17:03
  SPEAKER_01: the percentages of people that would figure out
• 17:06
  SPEAKER_01: that there were applications available and then where to buy them from,
• 17:09
  SPEAKER_01: you know, it was just diminishing.
• 17:10
  SPEAKER_01: And so it occurred to a bunch of us that on the General Magic,
• 17:14
  SPEAKER_01: Magic App interface,
• 17:15
  SPEAKER_01: you go outside and there's a street and on the street, there would be a store.
• 17:18
  SPEAKER_01: So like the That's why it's called the App Store
• 17:20
  SPEAKER_01: because General Magic's user interface was a bunch of buildings outside and
• 17:23
  SPEAKER_01: the store was the building that you went into to buy the apps.
• 17:26
  SPEAKER_01: And so that was just a business problem that occurred to us that the simplest way
• 17:31
  SPEAKER_01: to do this is on the device itself exposed to customers with a store metaphor,
• 17:37
  SPEAKER_01: which was we were really helped in our thinking
• 17:38
  SPEAKER_01: because there was already this metaphor to street.
• 17:40
  SPEAKER_01: And so at Philips,
• 17:42
  SPEAKER_01: Tony Fidel and I and Blake Gricorian and a number of others from General Magic went
• 17:46
  SPEAKER_01: to try to build And the idea there was, hey,
• 17:50
  SPEAKER_01: let's have it be so that not only can you buy an app one time,
• 17:54
  SPEAKER_01: but you could buy a subscription to an application and you could have try
• 17:57
  SPEAKER_01: before you buy.
• 18:00
  SPEAKER_01: And that was really the breakthrough there, again,
• 18:02
  SPEAKER_01: because we found that people didn't want to spend immediately $20 for something
• 18:08
  SPEAKER_01: that they hadn't tried before.
• 18:09
  SPEAKER_01: So we were just confronted every day with these developers calling us saying,
• 18:13
  SPEAKER_01: we're going broke, you need to help us make money.
• 18:15
  SPEAKER_01: And that led to these creative solutions.
• 18:17
  SPEAKER_02: I just want to give context for younger listeners or viewers out there.
• 18:21
  SPEAKER_02: I mean,
• 18:21
  SPEAKER_02: once upon a time you would buy floppy disks or you would buy CD-ROMs and have
• 18:27
  SPEAKER_02: to install software manually on a device.
• 18:30
  SPEAKER_02: I mean, that's unheard of today.
• 18:32
  SPEAKER_02: That you would have to engage in physical activity out in the real world of going
• 18:37
  SPEAKER_02: into a software store and schlepping the stuff home
• 18:40
  SPEAKER_02: and swapping out this floppy disk and then that one and waiting.
• 18:43
  SPEAKER_02: So the metaphor is apt because that's how people had to do it.
• 18:47
  SPEAKER_02: And it's really very hard for people to conceive who didn't actually experience that growing up.
• 18:54
  SPEAKER_01: In terms of the hijinks,
• 18:55
  SPEAKER_01: one of my favorites was Tom and I charaded as employees of the building next door
• 19:03
  SPEAKER_01: during their employee picnic.
• 19:04
  SPEAKER_01: And we were Tom and Steve from IT because we assumed that no one would recognize
• 19:09
  SPEAKER_01: the guys from IT.
• 19:11
  SPEAKER_01: And we did the lip sync contest and won.
• 19:13
  SPEAKER_01: It wasn't until after we had won that they realized that, in fact, we didn't work there.
• 19:18
  SPEAKER_02: We tried to claim that we were the night shift,
• 19:20
  SPEAKER_02: and that's why no one recognized us through that ruse.
• 19:24
  SPEAKER_02: I'm wondering,
• 19:25
  SPEAKER_02: just to get back to the second question I asked,
• 19:28
  SPEAKER_02: and you've talked about this a little bit, is there a methodology?
• 19:31
  SPEAKER_02: Is there a step-by-step to innovation?
• 19:33
  SPEAKER_01: Yeah, a lot of this stuff I learned from Tony Fidel,
• 19:35
  SPEAKER_01: who was my boss many times, including the iPod team.
• 19:39
  SPEAKER_01: And Tony also uses this notion that you should pick something that's like a medicine
• 19:42
  SPEAKER_01: to solve a real problem as opposed to a vitamin that's not...
• 20:04
  SPEAKER_01: So you want to map a real, real true need in their daily lives to your solution.
• 20:11
  SPEAKER_01: And then what you do is you say, okay, the person has whatever this need is.
• 20:15
  SPEAKER_01: They wake up in the morning.
• 20:16
  SPEAKER_01: How do they encounter,
• 20:18
  SPEAKER_01: how do they learn about your What's the experience for them to tell their friends
• 20:22
  SPEAKER_01: about it?
• 20:22
  SPEAKER_01: And are there incentives for them to tell their friends about it?
• 20:24
  SPEAKER_01: And that kind of thinking really...
• 20:34
  SPEAKER_01: Really helps you to think through the entire Indian experience and how
• 20:39
  SPEAKER_01: the marketing also relates to the product itself.
• 20:42
  SPEAKER_01: Another thing that Apple famously would do is Steve and
• 20:46
  SPEAKER_01: the product team would put together the marketing person with the engineer.
• 20:52
  SPEAKER_01: The engineer needed to describe what were the features that were the really exciting features.
• 20:56
  SPEAKER_01: And then the marketing person had to figure out how to make
• 20:58
  SPEAKER_01: a really compelling billboard campaign or TV campaign or magazine ad out of
• 21:03
  SPEAKER_01: that feature that would make sense and be compelling enough to drive someone to buy
• 21:07
  SPEAKER_01: the product.
• 21:08
  SPEAKER_01: And that was a threshold for knowing whether or not what was being built was going
• 21:13
  SPEAKER_01: to be successful or not.
• 21:14
  SPEAKER_02: So those are really, really powerful tools.
• 21:17
  SPEAKER_01: I also, even at General Magic, we did user testing.
• 21:20
  SPEAKER_01: And so bringing people in who are in your target audience and having them sit down
• 21:25
  SPEAKER_01: and have them use the product,
• 21:27
  SPEAKER_01: where you have someone who's kind of managing the session and you're recording it
• 21:31
  SPEAKER_01: or standing behind a mirror.
• 21:32
  SPEAKER_01: So you have to say things like, what do you see?
• 21:34
  SPEAKER_01: And what would you use this to do?
• 21:35
  SPEAKER_01: And what do you think you need to do next?
• 21:36
  SPEAKER_01: As opposed to leading them
• 21:54
  SPEAKER_01: And by having them kind of speak out loud, what are they seeing?
• 21:59
  SPEAKER_01: How are they thinking about it?
• 22:00
  SPEAKER_01: What do they think the thing does?
• 22:01
  SPEAKER_01: You also learn an enormous amount about what the product,
• 22:05
  SPEAKER_01: what the user interface is like,
• 22:06
  SPEAKER_01: what the functionality is like, what needs to be much more optimized.
• 22:10
  SPEAKER_01: Those are super powerful.
• 22:11
  SPEAKER_01: There's also two thinkers that I really like.
• 22:13
  SPEAKER_01: One's named Steve Blank who teaches at Stanford.
• 22:15
  SPEAKER_01: He's incredible.
• 22:17
  SPEAKER_01: He really kind of led the early thinking on customer discovery where you go
• 22:22
  SPEAKER_01: and you just before you even write code.
• 22:24
  SPEAKER_01: Yeah.
• 22:39
  SPEAKER_01: That stuff's really powerful, too.
• 23:06
  SPEAKER_01: And it's all about being close to the customer.
• 23:09
  SPEAKER_01: That's how you build great things is that you just have extreme sensitivity to customers.
• 23:14
  SPEAKER_02: First of all,
• 23:14
  SPEAKER_02: for the uninitiated, MVP, meaning minimal viable product, the concept of what is...
• 23:21
  SPEAKER_02: That's exactly right.
• 23:22
  SPEAKER_02: And one of the things that Steve Jobs used to famously say is that saying no is one
• 23:36
  SPEAKER_01: And so in the iPad.
• 23:51
  SPEAKER_01: But also leaving 10 to 20% of the engineering resources to solve the expert user problems.
• 23:55
  SPEAKER_01: So.
• 24:06
  SPEAKER_01: So the people who are the early adopters and who are the fanatics,
• 24:09
  SPEAKER_01: keeping them super happy at the same time.
• 24:12
  SPEAKER_01: So you introduce the features in each gradual release that are the obvious things
• 24:18
  SPEAKER_01: that attract new customers.
• 24:19
  SPEAKER_01: But you also reserve, again,
• 24:22
  SPEAKER_01: time where you're saying no to many,
• 24:23
  SPEAKER_01: many ideas and you're picking a few that are really,
• 24:25
  SPEAKER_01: really critical to keep the really avid fans happy.
• 24:29
  SPEAKER_01: I think that's also a really, really good idea.
• 24:31
  SPEAKER_01: Tony did that very actively at Nest.
• 24:34
  SPEAKER_03: Well, that's exactly what we didn't do at General Magic.
• 24:37
  SPEAKER_03: Yeah, right.
• 24:37
  SPEAKER_03: When we teach the case at the business schools, that's one of the identified failure modes.
• 24:41
  SPEAKER_03: But you talk, Steve,
• 24:43
  SPEAKER_03: about that people can imagine things that don't exist yet and then make them happen.
• 24:46
  SPEAKER_03: Well, what's an MVP for a moonshot?
• 24:49
  SPEAKER_03: What's an MVP for a vision like Mark Peratt's, which was the device, the network, the ecosystem?
• 24:55
  SPEAKER_03: What's an MVP for that?
• 24:57
  SPEAKER_01: Right, right.
• 24:58
  SPEAKER_01: Yeah.
• 24:58
  SPEAKER_01: Well, look, I mean, those kinds of visions occur very rarely.
• 25:03
  SPEAKER_01: The kinds of innovation vision I was talking about is you imagine a problem and it's
• 25:07
  SPEAKER_01: a daily life problem of many, many people and you set out to solve that problem.
• 25:12
  SPEAKER_01: Mark set out with a vision that was fundamental to everything.
• 25:16
  SPEAKER_01: It was fundamental to how we communicate,
• 25:19
  SPEAKER_01: fundamental to computers in our pockets, how all these things are interconnected.
• 25:22
  SPEAKER_01: I mean, it was just the depth and the prescience of the vision was just ridiculous.
• 25:28
  SPEAKER_01: It was just incredible.
• 25:30
  SPEAKER_01: So what is the MVP of General Magic?
• 25:32
  SPEAKER_01: I mean, what it proved to be was the Palm Pilot.
• 25:37
  SPEAKER_01: Palm Pilot turned out to be the MVP of General Magic,
• 25:39
  SPEAKER_01: which was, hey,
• 25:40
  SPEAKER_01: you want to get your contacts into this device that you can carry around with you.
• 25:44
  SPEAKER_01: It needs to sync effortlessly with your PC and get your contacts on it
• 25:47
  SPEAKER_01: so you can carry it around with your calendar.
• 25:49
  SPEAKER_01: And so it's got to be cheap, long battery life, fast and sync well.
• 25:54
  SPEAKER_01: And that was a great product.
• 25:56
  SPEAKER_01: Your listeners probably don't know this story,
• 25:58
  SPEAKER_01: but Jeff Hawkins,
• 25:59
  SPEAKER_01: originally, who is the CEO of Palm, he was a software developer for General Magic.
• 26:04
  SPEAKER_01: And he said, Oh, my God, I need to stop right now.
• 26:19
  SPEAKER_01: And then the next call,
• 26:20
  SPEAKER_01: which was maybe a month later,
• 26:22
  SPEAKER_01: he said,
• 26:22
  SPEAKER_01: okay, my board has decided to let me keep going, but I need to make my own devices.
• 26:26
  SPEAKER_01: And we said, you're crazy.
• 26:28
  SPEAKER_01: How are you ever going to compete with Sony and Motorola?
• 26:30
  SPEAKER_01: And he said, well, look, I see a path to doing it.
• 26:34
  SPEAKER_01: I have a sense of what's the right product to build that's narrowly constrained and
• 26:37
  SPEAKER_01: the board's willing to fund me to go for it.
• 26:39
  SPEAKER_01: So I'm just going to try.
• 26:41
  SPEAKER_01: And he ended up winning.
• 26:42
  SPEAKER_01: But the vision that we had was the right vision.
• 26:46
  SPEAKER_01: It was amazing.
• 26:48
  SPEAKER_01: It was the iPhone, and it was the cloud, and it was agents also, right?
• 26:52
  SPEAKER_01: All of the AI agents that we're facing today,
• 26:54
  SPEAKER_01: all the fundamental patents on those came from Jim White at General Magic
• 26:58
  SPEAKER_01: and Rich Miller.
• 26:59
  SPEAKER_01: And so the depth and the accuracy of the vision was amazing.
• 27:02
  SPEAKER_01: It was right.
• 27:03
  SPEAKER_01: It was intuitively correct when you heard it.
• 27:17
  SPEAKER_01: That, of course, that's what the future should look like.
• 27:19
  SPEAKER_01: So instead of trying to break off little incremental pieces,
• 27:23
  SPEAKER_01: we just tried to immediately jump 30 years ahead in the future and build it.
• 27:27
  SPEAKER_01: And it just wasn't possible.
• 27:28
  SPEAKER_02: Yeah, the old saw about you eat an elephant one bite at a time.
• 27:31
  SPEAKER_02: And also in technology, the idea of the killer app.
• 27:34
  SPEAKER_02: And we weren't looking to land and expand.
• 27:37
  SPEAKER_02: We were looking to take over the entire territory in one fell swoop.
• 27:41
  SPEAKER_01: And also, we were just so early.
• 27:43
  SPEAKER_01: I mean, I remembered vividly,
• 27:45
  SPEAKER_01: we went to CES,
• 27:46
  SPEAKER_01: which is this big consumer electronics show in the U.S., soon after we'd shipped.
• 27:50
  SPEAKER_01: And maybe, Tom, you were there.
• 27:51
  SPEAKER_01: We were demoing the product in the Sony booth to everybody.
• 27:54
  SPEAKER_01: And everybody kept saying,
• 27:55
  SPEAKER_01: well, where's And we said, oh, you plug the phone line into the side.
• 27:58
  SPEAKER_01: And they said, no, no, no, I want a cellular phone inside of that thing.
• 28:02
  SPEAKER_01: And they were right.
• 28:03
  SPEAKER_01: And we came back from that show knowing that that's what we had to build.
• 28:07
  SPEAKER_01: But that was before digital cellular existed.
• 28:09
  SPEAKER_01: There was no, it was just analog cellular.
• 28:12
  SPEAKER_01: So even just on the product that we knew to build that was the iPhone,
• 28:17
  SPEAKER_01: we were 15 years early, even on just the telco infrastructure side.
• 28:22
  SPEAKER_01: So we were just wildly early.
• 28:24
  SPEAKER_02: I want to come back to something you said about innovation and product development
• 28:28
  SPEAKER_02: and jobs,
• 28:29
  SPEAKER_02: thought experiment of a day in the life of somebody who's a customer for something
• 28:34
  SPEAKER_02: that doesn't exist yet.
• 28:35
  SPEAKER_02: And part of that is Once they're using it,
• 28:38
  SPEAKER_02: how they communicate to other people, how other people find out about it.
• 28:42
  SPEAKER_02: What I hear you saying is that as an innovator,
• 28:45
  SPEAKER_02: it's not a technical endeavor,
• 28:47
  SPEAKER_02: certainly not solely and maybe even not primarily, but it's a social endeavor.
• 28:52
  SPEAKER_02: That you were engineering an experience,
• 28:55
  SPEAKER_02: you're engineering something that's going to be contagious socially.
• 28:58
  SPEAKER_02: Has that been your experience that that's a key component of it?
• 29:01
  SPEAKER_01: Yeah, yeah.
• 29:02
  SPEAKER_01: I mean, on the technical side, it helps if you see opportunities early.
• 29:07
  SPEAKER_01: So it helps to know that because of some advancement in hardware technology or
• 29:13
  SPEAKER_01: a new technology,
• 29:14
  SPEAKER_01: Advancement in software technology that enables something new, like,
• 29:18
  SPEAKER_01: you know,
• 29:18
  SPEAKER_01: the ability to determine the location of the phone,
• 29:20
  SPEAKER_01: having early visibility of those technological enablers,
• 29:25
  SPEAKER_01: and then you creatively leap from there to build a platform.
• 29:29
  SPEAKER_01: So you can't underrate that.
• 29:31
  SPEAKER_01: But certainly having it be really effortless for people to try the product
• 29:36
  SPEAKER_01: and discover the product and then include their friends to join.
• 29:40
  SPEAKER_01: One of the lessons at Facebook,
• 29:42
  SPEAKER_01: there's a brilliant person who's still Who's
• 29:44
  SPEAKER_01: the chief marketing officer named Alex Schultz.
• 29:46
  SPEAKER_01: He just wrote a very good book called Click that I highly recommend people read.
• 29:51
  SPEAKER_01: His point of view was the primary determination of success of people
• 29:54
  SPEAKER_01: on Facebook was could you get them connected to a certain number of friends
• 29:58
  SPEAKER_01: as quickly as possible?
• 29:59
  SPEAKER_01: Because then they started to see utility and they started to see things from their friends.
• 30:04
  SPEAKER_01: They started to like what the friends were saying.
• 30:06
  SPEAKER_01: That then led the friends in turn to like whatever they had posted it.
• 30:10
  SPEAKER_01: And that started the ball rolling.
• 30:12
  SPEAKER_01: The notion that you're rewarding customers for behavior,
• 30:15
  SPEAKER_01: if you look at all of the apps for learning new languages,
• 30:19
  SPEAKER_01: the apps for sports,
• 30:20
  SPEAKER_01: for health,
• 30:21
  SPEAKER_01: those are all,
• 30:22
  SPEAKER_01: they all have these sort of gaming dynamics that encourage you to continue to use
• 30:27
  SPEAKER_01: the product.
• 30:27
  SPEAKER_01: And those are really powerful.
• 30:29
  SPEAKER_01: So there's a lot of the social dynamic and the sort of psychological dynamic
• 30:33
  SPEAKER_01: in applications is important.
• 30:36
  SPEAKER_03: When you were building Audible...
• 30:38
  SPEAKER_03: What did you think the market was going to be?
• 30:39
  SPEAKER_03: Did you think people listened to stuff in cars?
• 30:41
  SPEAKER_03: Did you already see it as mobile?
• 30:43
  SPEAKER_01: Guy Story and Don Katz,
• 30:45
  SPEAKER_01: who started Audible,
• 30:46
  SPEAKER_01: had this amazing idea that when your hands are busy, that your mind is free.
• 30:51
  SPEAKER_01: And Guy was an early, early audio, really deep audio technological engineer from Bell Labs.
• 30:58
  SPEAKER_01: So he was incredibly good at making speech sound great in tiny amounts of memory.
• 31:04
  SPEAKER_01: And so that was their innovation was they built this tiny little hardware player
• 31:10
  SPEAKER_01: that you would plug into the cigarette lighter jack of your car and you play
• 31:15
  SPEAKER_01: the audio when you were on the way to work.
• 31:17
  SPEAKER_01: My contribution was I saw that in a bookstore.
• 31:20
  SPEAKER_01: And Tony and I at Philips had been building a precursor of the iPhone
• 31:25
  SPEAKER_01: that was called the Nino, which is a little handheld device.
• 31:28
  SPEAKER_01: And it had a headphone jack in the bottom.
• 31:31
  SPEAKER_01: And so I contacted Guy, who's the CTO, and said, hey, you don't want to be selling these.
• 31:35
  SPEAKER_01: Tony's hardware players in bookshops.
• 31:37
  SPEAKER_01: You want me to write you an app that runs in this device that Tony's building
• 31:41
  SPEAKER_01: at Philips, and you have a software business.
• 31:44
  SPEAKER_01: And so he listened to my crazy idea,
• 31:46
  SPEAKER_01: and that was the Audible Player,
• 31:49
  SPEAKER_01: which was the first commercial audio app ever on a mobile device.
• 31:55
  SPEAKER_01: And those guys are just wonderful to work with.
• 31:58
  SPEAKER_03: I miss the opportunity to work with them.
• 32:00
  SPEAKER_03: It was 8 o'clock at night,
• 32:01
  SPEAKER_03: one night in my office, just as the wave of the first bubble was cresting.
• 32:05
  SPEAKER_03: And we were so busy at the law firm.
• 32:07
  SPEAKER_03: And they said, we're doing this thing.
• 32:09
  SPEAKER_03: Do you want to represent us?
• 32:11
  SPEAKER_03: I said, we're just too busy.
• 32:12
  SPEAKER_03: Sorry.
• 32:12
  SPEAKER_01: Yeah.
• 32:14
  SPEAKER_01: In fact,
• 32:14
  SPEAKER_01: I don't know if you guys know this,
• 32:16
  SPEAKER_01: but the user interface of the Audible player,
• 32:19
  SPEAKER_01: I was looking for something that was really modern and really,
• 32:22
  SPEAKER_01: really beautiful and kind of curved.
• 32:24
  SPEAKER_01: And Zarkov Dragonich,
• 32:25
  SPEAKER_01: who's an engineer that we used to work with,
• 32:27
  SPEAKER_01: had just I bought a Porsche 911 and the dashboard of the Porsche 911 was,
• 32:32
  SPEAKER_01: I just stole the design from the dashboard of the 911 and that was the design
• 32:37
  SPEAKER_01: the auto player.
• 32:38
  SPEAKER_01: And then Bill Gates loved the app because it was a sign The Windows products,
• 32:44
  SPEAKER_01: they were these early Windows mobile devices, wasn't just for work.
• 32:49
  SPEAKER_01: It was also for fun.
• 32:50
  SPEAKER_01: And so there was a full page screenshot of the UI in the Microsoft Annual Report that year.
• 32:56
  SPEAKER_01: So it all came from Xarco's 911.
• 32:58
  SPEAKER_02: Steve, I wanted to ask you,
• 33:00
  SPEAKER_02: not from a business standpoint,
• 33:01
  SPEAKER_02: but from the standpoint of its social impact or political or cultural impact,
• 33:06
  SPEAKER_02: are there things that keep you up at night about AI?
• 33:08
  SPEAKER_02: And if so, what are those?
• 33:10
  SPEAKER_01: Yeah, I talked already about my concerns about the really,
• 33:13
  SPEAKER_01: really powerful So that's the first thing I'm also concerned about.
• 33:18
  SPEAKER_01: We talked a bit about Africa.
• 33:19
  SPEAKER_01: At Orange,
• 33:20
  SPEAKER_01: one of the things I'm really proud of is that we figured out how
• 33:23
  SPEAKER_01: to customize existing language models to recognize a bunch
• 33:27
  SPEAKER_01: of regional African languages.
• 33:28
  SPEAKER_01: So Wolof, for example, is a regional language spoken in Senegal by about 30 million people.
• 33:34
  SPEAKER_01: And no AI model understands Wolof today.
• 33:37
  SPEAKER_01: And so we built models and open source them that do.
• 33:40
  SPEAKER_01: And for me,
• 33:41
  SPEAKER_01: that was about getting ahead of the problem of the massive digital divide
• 33:46
  SPEAKER_01: between not only the have and have-nots in terms of people that can afford
• 33:50
  SPEAKER_01: and not afford,
• 33:51
  SPEAKER_01: but these are people who can be just completely locked out
• 33:53
  SPEAKER_01: of using these systems just because...
• 33:58
  SPEAKER_01: So I think we have to anticipate societal problems where the gap between
• 34:03
  SPEAKER_01: the productivity gap,
• 34:05
  SPEAKER_01: people that are able to have access to the most powerful tools and those
• 34:09
  SPEAKER_01: that don't is going to be large.
• 34:11
  SPEAKER_01: And that's going to be true in the U.S.,
• 34:12
  SPEAKER_01: but it's especially going to be true between markets like the U.S.
• 34:15
  SPEAKER_01: and Western Europe and developing countries like in West Africa, for example.
• 34:19
  SPEAKER_01: So that's something that I'm concerned about.
• 34:22
  SPEAKER_01: As you think ahead beyond AI into robotics,
• 34:25
  SPEAKER_01: I think that problem becomes even greater because if you think about the amount
• 34:29
  SPEAKER_01: of leisure time that people have,
• 34:32
  SPEAKER_01: you're going to buy a car and then you're going to buy a refrigerator and
• 34:36
  SPEAKER_01: then you can buy a TV and then you can buy a robot and maybe the robot becomes for
• 34:40
  SPEAKER_01: the TV.
• 34:41
  SPEAKER_01: And so the people that can afford robots to help them in their daily lives,
• 34:45
  SPEAKER_01: they're going to have an extra couple of hours a day for leisure.
• 34:49
  SPEAKER_01: And so I think there too, we need to be very careful about the gap.
• 34:52
  SPEAKER_01: The challenge there is that a lot of people think,
• 34:55
  SPEAKER_01: well, the notion is that we'll have universal basic income and...
• 34:59
  SPEAKER_01: We'll have additional tax revenue based on the incremental value generated
• 35:03
  SPEAKER_01: by these people using these AI systems, which I'd like to believe will happen.
• 35:09
  SPEAKER_01: But historically,
• 35:10
  SPEAKER_01: government's ability to effectively
• 35:12
  SPEAKER_01: and fairly redistribute income is pretty limited.
• 35:15
  SPEAKER_01: Or when it does occur...
• 35:18
  SPEAKER_01: And then we talked before about...
• 35:40
  SPEAKER_01: What do you do about people that don't want to use these tools?
• 35:44
  SPEAKER_01: For me, a good metaphor is a business run on Excel versus being run on an abacus.
• 35:49
  SPEAKER_01: And I think it's at least that big of a gap.
• 35:51
  SPEAKER_01: And so what do you do with the people in a company or in society
• 35:55
  SPEAKER_01: that they like their abacus and they don't want to move to Excel?
• 35:57
  SPEAKER_01: It's going to be a challenge.
• 35:59
  SPEAKER_01: One of the things that gets me most excited about AI is the fact
• 36:02
  SPEAKER_01: that it's this natural language interaction with a computer.
• 36:06
  SPEAKER_01: So for me, that's a total game changer.
• 36:08
  SPEAKER_01: Just speaking plain language to a computer and having an understanding
• 36:12
  SPEAKER_01: a very large percentage of what it is that you want to try to do.
• 36:14
  SPEAKER_01: And then the computer's fairly good at programming to generate that thing for you.
• 36:18
  SPEAKER_01: And that is massive, massive breakthrough.
• 36:21
  SPEAKER_01: So I think there are going to be a bunch of challenges,
• 36:24
  SPEAKER_01: but I think the fact that the primary interaction method
• 36:28
  SPEAKER_01: with these systems is just talking or writing,
• 36:31
  SPEAKER_01: which is super natural for people to do,
• 36:32
  SPEAKER_01: I think that that's going to mitigate some of the challenges,
• 36:36
  SPEAKER_01: but the disparity problem is still going to be big.
• 36:38
  SPEAKER_03: Two things.
• 36:39
  SPEAKER_03: One thing,
• 36:40
  SPEAKER_03: what's the right platform for the optimal deployment in terms of accessibility
• 36:44
  SPEAKER_03: and usability?
• 36:45
  SPEAKER_03: But also right now,
• 36:46
  SPEAKER_03: most of the benefits of the AI revolution are redounding to a very small handful
• 36:51
  SPEAKER_03: of companies run by tech bros who are either libertarian or fascist.
• 36:58
  SPEAKER_03: That's my personal opinion.
• 36:59
  SPEAKER_03: Do ideas like a token tax have any traction with you as a way of trying to, again,
• 37:04
  SPEAKER_03: redistribute the economic benefits of this gigantic multi-trillion dollar investment
• 37:10
  SPEAKER_03: in AI infrastructure?
• 37:11
  SPEAKER_03: Yeah.
• 37:12
  SPEAKER_03: You know, what it takes to build and train.
• 37:27
  SPEAKER_01: So I love that scenario for skiing.
• 37:29
  SPEAKER_01: So I was wearing my glasses skiing recently with my daughter and
• 37:55
  SPEAKER_01: The idea that I could easily make a video just of what I'm seeing and I can just ski
• 37:58
  SPEAKER_01: behind her.
• 37:59
  SPEAKER_01: It plays music for me.
• 38:02
  SPEAKER_01: It gives me directions if I get lost.
• 38:04
  SPEAKER_01: Those use cases, again,
• 38:06
  SPEAKER_01: very similar to the Audible case where your hands are busy,
• 38:09
  SPEAKER_01: but your mind is Those are powerful.
• 38:11
  SPEAKER_01: And that also applies,
• 38:12
  SPEAKER_01: I think,
• 38:12
  SPEAKER_01: to sports like tennis and basketball and golf and other kinds
• 38:16
  SPEAKER_01: of exercise like running and so on.
• 38:18
  SPEAKER_01: So I think there's a really interesting market segment for smart glasses,
• 38:23
  SPEAKER_01: but the vast majority of the world will start with a phone and
• 38:26
  SPEAKER_01: then some people will also have glasses or maybe use glasses instead of a phone
• 38:31
  SPEAKER_01: in certain use cases like sports.
• 38:33
  SPEAKER_01: But I think phone is the vast, still the big winner there.
• 38:36
  SPEAKER_01: Yeah.
• 38:36
  SPEAKER_01: In terms of how we redistribute income,
• 38:39
  SPEAKER_01: I think it falls to the local governments,
• 38:42
  SPEAKER_01: especially the American government,
• 38:43
  SPEAKER_01: to make sure that the companies that are generating all
• 38:46
  SPEAKER_01: of these revenues are taxed appropriately,
• 38:48
  SPEAKER_01: and then they help redistribute the income.
• 38:52
  SPEAKER_01: I think the challenge becomes the disparity in access between people in the U.S.,
• 38:59
  SPEAKER_01: even poor people in the U.S. and poor people in developing countries is massive.
• 39:04
  SPEAKER_01: So in Congo, for example, most people...
• 39:06
  SPEAKER_01: And so I think that's another challenge in this,
• 39:09
  SPEAKER_01: what I see as a massive disparity in the world.
• 39:34
  SPEAKER_01: So whether it's tax or whether it's just general taxation,
• 39:39
  SPEAKER_01: I think that the challenge is You need
• 39:41
  SPEAKER_01: to have these corporations appropriately attached.
• 39:43
  SPEAKER_02: You mentioned that in France,
• 39:44
  SPEAKER_02: there's lifetime employment, whereas in the United States, there isn't.
• 39:48
  SPEAKER_02: And presumably some countries are better positioned to slow the economic disruption
• 39:55
  SPEAKER_02: on the workforce,
• 39:55
  SPEAKER_02: at least,
• 39:56
  SPEAKER_02: that AI will have,
• 39:58
  SPEAKER_02: whereas other countries are closer to a kind of Wild West,
• 40:01
  SPEAKER_02: where that kind of disruption is just going to be very rapid without much of
• 40:07
  SPEAKER_02: a soft landing for millions of people.
• 40:09
  SPEAKER_01: I mean, the surprising thing is if you compare the tax rates of someone who lives
• 40:14
  SPEAKER_01: in Berkeley,
• 40:14
  SPEAKER_01: California,
• 40:15
  SPEAKER_01: for example,
• 40:16
  SPEAKER_01: compared to somebody who lives in France, it's really not that different.
• 40:19
  SPEAKER_01: But in France, you've got amazing public health care.
• 40:22
  SPEAKER_01: You've got amazing public education.
• 40:24
  SPEAKER_01: The parks are all perfect.
• 40:26
  SPEAKER_01: You've got all this incredible social activity.
• 40:28
  SPEAKER_01: You've got great public transport.
• 40:30
  SPEAKER_01: Now they strike regularly.
• 40:31
  SPEAKER_01: So that's not universally available.
• 40:34
  SPEAKER_01: And so I think you also have to think a bit about what are the trade-offs
• 40:38
  SPEAKER_01: that you want to make in your society.
• 40:40
  SPEAKER_01: Now, France, the government is way too big.
• 40:42
  SPEAKER_01: There's way too much regulation.
• 40:44
  SPEAKER_01: The amount of bureaucracy and every little thing you want to try to do is extreme.
• 40:47
  SPEAKER_01: But the quality of life...
• 40:49
  SPEAKER_01: And so, yeah,
• 40:50
  SPEAKER_01: I think in countries like France and Spain and Italy and certainly the Nordics,
• 40:53
  SPEAKER_01: the nets that are provided by the government will help with the transition.
• 40:56
  SPEAKER_01: It's going to be more difficult in the USA.
• 41:10
  SPEAKER_02: Question out of left field,
• 41:11
  SPEAKER_02: what have you observed in terms of how AI influences the way people think or
• 41:16
  SPEAKER_02: the way they problem solve or just how they approach their work in general?
• 41:25
  SPEAKER_01: And so one of my favorite things to do is to say,
• 41:31
  SPEAKER_01: here's a problem I have,
• 41:32
  SPEAKER_01: look for the leading research on this problem,
• 41:35
  SPEAKER_01: give me competing points of view on how to solve the problem,
• 41:38
  SPEAKER_01: and then I'll often have multiple differences.
• 41:40
  SPEAKER_01: Yeah, absolutely.
• 41:55
  SPEAKER_01: And so the notion that you can very rapidly try new ideas, whereas before...
• 42:21
  SPEAKER_01: So, so the, the
• 42:36
  SPEAKER_01: The pace of innovation
• 42:37
  SPEAKER_01: in just everyday software development is getting really radically changed.
• 42:42
  SPEAKER_01: And many,
• 42:43
  SPEAKER_01: many more people who have no idea how to program a computer are going
• 42:47
  SPEAKER_01: to become programmers.
• 43:07
  SPEAKER_01: What we're beginning to see is,
• 43:09
  SPEAKER_01: it's still buggy,
• 43:10
  SPEAKER_01: but the notion that you can just have natural language interaction with the data of
• 43:14
  SPEAKER_01: the company directly means that many,
• 43:17
  SPEAKER_01: many people in the company are being data-driven,
• 43:19
  SPEAKER_01: and that's also going to be a massive shift for companies to be a lot more efficient
• 43:24
  SPEAKER_01: and effective.
• 43:24
  SPEAKER_03: Does it create new jobs in a way that is actually likely to offset the job destruction?
• 43:30
  SPEAKER_03: That's my perspective.
• 43:31
  SPEAKER_03: I think we're going to have...
• 43:45
  SPEAKER_01: Way more people programming, like I said before, than we have today.
• 43:49
  SPEAKER_01: We're going to have much more software generated.
• 43:51
  SPEAKER_01: If you look at the world's best companies,
• 43:53
  SPEAKER_01: if you look at a company like Apple or a company like Amazon,
• 43:56
  SPEAKER_01: and you look at the level of sophistication of their customer support
• 44:00
  SPEAKER_01: and their website,
• 44:03
  SPEAKER_01: And the retail stores in the case of Apple,
• 44:05
  SPEAKER_01: and you compare with almost any other business in the world,
• 44:07
  SPEAKER_01: there's just an enormous gap.
• 44:10
  SPEAKER_01: And so if AI enables businesses to have customer experiences,
• 44:16
  SPEAKER_01: whether they be digital or in retail,
• 44:18
  SPEAKER_01: closer to those kinds of experiences, I think that's a huge win.
• 44:23
  SPEAKER_01: Now, if you think out 50 years now,
• 44:26
  SPEAKER_01: And if you believe,
• 44:28
  SPEAKER_01: like I do,
• 44:28
  SPEAKER_01: that this is like an agrarian era to industrial era level transformation,
• 44:34
  SPEAKER_01: it's hard to say.
• 44:36
  SPEAKER_01: Will people have a lot more leisure time?
• 44:38
  SPEAKER_01: I think the people that live in these economies where
• 44:41
  SPEAKER_01: the wealth is better redistributed and they have access to these kinds of tools in
• 44:45
  SPEAKER_01: an affordable way,
• 44:46
  SPEAKER_01: it's very possible that you're going to have people with a lot more leisure time.
• 44:50
  SPEAKER_01: I mean, maybe this is kind of the next level state of human evolution is that the amount
• 44:54
  SPEAKER_01: that we actually have to work ourselves is reduced significantly.
• 44:58
  SPEAKER_01: And that creates a lot more free time for people to pursue hobbies and things.
• 45:02
  SPEAKER_01: That's the upside potential.
• 45:05
  SPEAKER_01: Downside potential is that, like you were asking, Mike, a bunch of people's jobs are disrupted.
• 45:09
  SPEAKER_01: There aren't job replacements or ones that are interesting to people.
• 45:13
  SPEAKER_01: The income redistribution with the governments doesn't work.
• 45:16
  SPEAKER_01: And then you just have this whole set of people that can't find something to do.
• 45:20
  SPEAKER_01: But I think it's the former.
• 45:22
  SPEAKER_01: I think there's just a bunch of new jobs created.
• 45:25
  SPEAKER_01: I mean, think of how many robot repairmen there are going to be in 30 years, right?
• 45:31
  SPEAKER_01: There's going to be robot repairmen everywhere.
• 45:34
  SPEAKER_03: It's not plumbing or HVAC, it's robot repairmen.
• 45:39
  SPEAKER_01: So like we had agrarian to industrial,
• 45:43
  SPEAKER_01: you had all these new specializations of jobs that were created as opposed to being
• 45:49
  SPEAKER_01: a farmer or a soldier or a priest.
• 45:51
  SPEAKER_01: Suddenly there were thousands of these job specializations.
• 45:55
  SPEAKER_01: I assume that's going to But I think the disparity gap in terms of all
• 46:02
  SPEAKER_01: of this stuff is the problem.
• 46:04
  SPEAKER_01: And what complicates it is that we're not just talking about Western Europe
• 46:08
  SPEAKER_01: and China and the U.S.
• 46:10
  SPEAKER_01: We're talking about countries where people have the average,
• 46:14
  SPEAKER_01: the birth rate is seven or the ability,
• 46:17
  SPEAKER_01: the disposable income is five euros per month.
• 46:20
  SPEAKER_01: I mean, some of these countries are very, very poor with enormously high birth rates.
• 46:24
  SPEAKER_01: So that's what concerns me.
• 46:26
  SPEAKER_02: Steve, I'd like to do a lightning true or false round with you.
• 46:30
  SPEAKER_02: True or false, you wrote the world's first mobile web browser.
• 46:34
  SPEAKER_01: I was the product manager on the world's first mobile browser.
• 46:37
  SPEAKER_01: It was written by a guy named Jeff Bone in Austin,
• 46:41
  SPEAKER_01: Texas, who He was a software engineer at a company called Active Paper.
• 46:44
  SPEAKER_01: And then the second one was written by Don Wood at General Magic,
• 46:49
  SPEAKER_01: who wrote Colossal Cave, which was the first ever computer game.
• 46:54
  SPEAKER_01: So Don Wood is also a man of many firsts.
• 46:59
  SPEAKER_02: True or false, you were a prominent member of the iPod and then the iPhone development teams.
• 47:05
  SPEAKER_01: Yeah, that's right.
• 47:06
  SPEAKER_01: Tony Fidel hired me into the early iPod team,
• 47:10
  SPEAKER_01: and my job and my team's job was to add Windows connectivity.
• 47:14
  SPEAKER_01: And we were in the audience when Steve was asked,
• 47:17
  SPEAKER_01: hey, are you ever going to have Windows connectivity?
• 47:19
  SPEAKER_01: He said, absolutely never.
• 47:20
  SPEAKER_01: Right.
• 47:21
  SPEAKER_01: What happened was,
• 47:22
  SPEAKER_01: so I was on the team for the first couple of devices,
• 47:25
  SPEAKER_01: and then I had an opportunity to go run mobile for Kodak and build
• 47:30
  SPEAKER_01: the very first applications that ran on phones for photography.
• 47:33
  SPEAKER_01: But everybody else on my team went into the iPhone team.
• 47:37
  SPEAKER_01: Tony hired me again later to help with connecting the iPhone to cars.
• 47:42
  SPEAKER_01: So the initial experience of when you plugged your phone into the car, I worked on.
• 47:47
  SPEAKER_03: In other words, CarPlay.
• 47:48
  SPEAKER_03: You did CarPlay.
• 47:49
  SPEAKER_01: Yeah, the first version of CarPlay.
• 47:51
  SPEAKER_01: So initially the way CarPlay worked was that we defined the software for
• 47:56
  SPEAKER_01: the car manufacturers to manage the display and the phone would talk to the software
• 48:02
  SPEAKER_01: in the car manufacturers system.
• 48:04
  SPEAKER_01: And that proved to be such a mess that instead what we did was the iPhone
• 48:08
  SPEAKER_01: then just drew all the user interface on the LCD of the,
• 48:12
  SPEAKER_01: we just completely took over the UI of the car, which was a much easier solution.
• 48:16
  SPEAKER_02: But that took some persuading.
• 48:19
  SPEAKER_02: I'm wondering what you learned at General Magic,
• 48:22
  SPEAKER_02: especially with respect to things that didn't go right,
• 48:26
  SPEAKER_02: that informs your work today as an innovator.
• 48:28
  SPEAKER_01: Yeah, yeah.
• 48:29
  SPEAKER_01: I mean, we talked about this earlier that being really close to customers makes
• 48:34
  SPEAKER_01: an enormous difference.
• 48:35
  SPEAKER_01: And the earlier you're exposed to customers, the better.
• 48:39
  SPEAKER_01: And you don't have to even have software to do that very, very effectively.
• 48:43
  SPEAKER_01: I think that was the lesson,
• 48:45
  SPEAKER_01: the hard lesson that I learned that we'd worked for years in secret.
• 48:49
  SPEAKER_01: And then as soon as we went to show the product to people at trade shows,
• 48:53
  SPEAKER_01: their feedback was So clear in terms of what they wanted to see that that was
• 49:00
  SPEAKER_01: a hard learned lesson.
• 49:01
  SPEAKER_01: And so one of the things that I do now is I try to really gather customer input in
• 49:06
  SPEAKER_01: as many different ways as I can before we spend very much money
• 49:10
  SPEAKER_01: on actual development.
• 49:12
  SPEAKER_01: I think also taking things in iterative steps where you really want
• 49:17
  SPEAKER_01: to understand what are the features that are going to be really appealing for people
• 49:20
  SPEAKER_01: and focus on just a few things that you do incredibly well.
• 49:24
  SPEAKER_01: And so you have this wonderful,
• 49:26
  SPEAKER_01: joyous Surprise experience on these key features that delight your customer
• 49:32
  SPEAKER_01: and those are the messages that you have your marketing communicate about and then
• 49:37
  SPEAKER_01: the word of mouth as they describe how exciting using the product was
• 49:41
  SPEAKER_01: that is amplified by the marketing and then you're off to the races.
• 49:45
  SPEAKER_01: But often it's hard to know what really is going to resonate with customers.
• 49:50
  SPEAKER_01: And so you have to be really willing to be surprised.
• 49:54
  SPEAKER_01: I had a software company that failed.
• 49:55
  SPEAKER_01: One of the reasons was once we were confronted with the fact that it wasn't software
• 50:00
  SPEAKER_01: that we should be building, we should be building hardware.
• 50:03
  SPEAKER_01: That transition was just so hard for me to grasp that we didn't go for it.
• 50:09
  SPEAKER_01: And so I think you have to be ready to be really surprised about what works
• 50:13
  SPEAKER_01: and what doesn't work.
• 50:14
  SPEAKER_03: Okay.
• 50:15
  SPEAKER_03: I'd like to keep going for another couple of hours, but we got to let you go.
• 50:18
  SPEAKER_03: It's late.
• 50:18
  SPEAKER_03: It's late in Paris.
• 50:19
  SPEAKER_03: You bet.
• 50:20
  SPEAKER_03: Thank you, Sue.
• 50:21
  SPEAKER_03: That was wonderful.
• 50:22
  SPEAKER_00: Thank you for listening to General Magic & The Soul of Tech podcast.
• 50:26
  SPEAKER_00: Tune into our Substack for more interviews and articles
• 50:29
  SPEAKER_00: at spellboundproductions.substack.com.