vtt/episode_030_small.vtt

WEBVTT

00:00.000 --> 00:03.440
 The following is a conversation with Kevin Scott,

00:03.440 --> 00:06.080
 the CTO of Microsoft.

00:06.080 --> 00:08.560
 Before that, he was the senior vice president

00:08.560 --> 00:11.080
 of engineering and operations at LinkedIn,

00:11.080 --> 00:13.520
 and before that, he oversaw mobile ads

00:13.520 --> 00:14.960
 engineering at Google.

00:15.960 --> 00:19.000
 He also has a podcast called Behind the Tech

00:19.000 --> 00:21.880
 with Kevin Scott, which I'm a fan of.

00:21.880 --> 00:24.280
 This was a fun and wide ranging conversation

00:24.280 --> 00:26.680
 that covered many aspects of computing.

00:26.680 --> 00:28.840
 It happened over a month ago,

00:28.840 --> 00:31.000
 before the announcement of Microsoft's investment

00:31.000 --> 00:34.440
 OpenAI that a few people have asked me about.

00:34.440 --> 00:38.120
 I'm sure there'll be one or two people in the future

00:38.120 --> 00:41.400
 that'll talk with me about the impact of that investment.

00:42.280 --> 00:45.440
 This is the Artificial Intelligence podcast.

00:45.440 --> 00:47.680
 If you enjoy it, subscribe on YouTube,

00:47.680 --> 00:49.440
 give it five stars on iTunes,

00:49.440 --> 00:50.960
 support it on a Patreon,

00:50.960 --> 00:53.000
 or simply connect with me on Twitter,

00:53.000 --> 00:57.680
 at Lex Freedman, spelled FRIDMAM.

00:57.680 --> 00:59.240
 And I'd like to give a special thank you

00:59.240 --> 01:01.960
 to Tom and Elanti Bighausen

01:01.960 --> 01:04.600
 for their support of the podcast on Patreon.

01:04.600 --> 01:06.080
 Thanks Tom and Elanti.

01:06.080 --> 01:08.400
 Hope I didn't mess up your last name too bad.

01:08.400 --> 01:10.520
 Your support means a lot,

01:10.520 --> 01:13.480
 and inspires me to keep this series going.

01:13.480 --> 01:18.160
 And now, here's my conversation with Kevin Scott.

01:18.160 --> 01:20.760
 You've described yourself as a kid in a candy store

01:20.760 --> 01:23.000
 at Microsoft because of all the interesting projects

01:23.000 --> 01:24.200
 that are going on.

01:24.200 --> 01:28.000
 Can you try to do the impossible task

01:28.000 --> 01:31.760
 and give a brief whirlwind view

01:31.760 --> 01:34.520
 of all the spaces that Microsoft is working in?

01:35.520 --> 01:37.440
 Both research and product.

01:37.440 --> 01:42.440
 If you include research, it becomes even more difficult.

01:46.480 --> 01:48.880
 So, I think broadly speaking,

01:48.880 --> 01:53.720
 Microsoft's product portfolio includes everything

01:53.720 --> 01:56.920
 from big cloud business,

01:56.920 --> 01:59.360
 like a big set of SaaS services.

01:59.360 --> 02:01.720
 We have sort of the original,

02:01.720 --> 02:05.560
 or like some of what are among the original

02:05.560 --> 02:09.640
 productivity software products that everybody uses.

02:09.640 --> 02:11.200
 We have an operating system business.

02:11.200 --> 02:13.560
 We have a hardware business

02:13.560 --> 02:17.240
 where we make everything from computer mice

02:17.240 --> 02:20.760
 and headphones to high end,

02:20.760 --> 02:23.520
 high end personal computers and laptops.

02:23.520 --> 02:27.680
 We have a fairly broad ranging research group

02:27.680 --> 02:29.680
 where we have people doing everything

02:29.680 --> 02:31.880
 from economics research.

02:31.880 --> 02:35.920
 So, there's this really smart young economist,

02:35.920 --> 02:39.760
 Glenn Weil, who like my group works with a lot,

02:39.760 --> 02:42.880
 who's doing this research on these things

02:42.880 --> 02:45.120
 called radical markets.

02:45.120 --> 02:48.120
 Like he's written an entire technical book

02:48.120 --> 02:51.120
 about this whole notion of radical markets.

02:51.120 --> 02:53.520
 So, like the research group sort of spans from that

02:53.520 --> 02:56.840
 to human computer interaction, to artificial intelligence.

02:56.840 --> 03:01.040
 And we have GitHub, we have LinkedIn.

03:01.040 --> 03:05.800
 We have a search advertising and news business

03:05.800 --> 03:07.360
 and like probably a bunch of stuff

03:07.360 --> 03:11.240
 that I'm embarrassingly not recounting in this list.

03:11.240 --> 03:12.920
 On gaming to Xbox and so on, right?

03:12.920 --> 03:14.120
 Yeah, gaming for sure.

03:14.120 --> 03:17.320
 Like I was having a super fun conversation

03:17.320 --> 03:19.520
 this morning with Phil Spencer.

03:19.520 --> 03:21.280
 So, when I was in college,

03:21.280 --> 03:25.560
 there was this game that Lucas Arts made

03:25.560 --> 03:27.600
 called Day of the Tentacle,

03:27.600 --> 03:30.160
 that my friends and I played forever.

03:30.160 --> 03:33.920
 And like we're doing some interesting collaboration now

03:33.920 --> 03:37.920
 with the folks who made Day of the Tentacle.

03:37.920 --> 03:40.840
 And I was like completely nerding out with Tim Schaeffer,

03:40.840 --> 03:43.880
 like the guy who wrote Day of the Tentacle this morning,

03:43.880 --> 03:45.840
 just a complete fanboy,

03:45.840 --> 03:49.880
 which you know, sort of it like happens a lot.

03:49.880 --> 03:53.320
 Like, you know, Microsoft has been doing so much stuff

03:53.320 --> 03:56.000
 at such breadth for such a long period of time

03:56.000 --> 03:59.680
 that, you know, like being CTO,

03:59.680 --> 04:02.200
 like most of the time my job is very, very serious

04:02.200 --> 04:05.640
 and sometimes that like I get caught up

04:05.640 --> 04:09.200
 in like how amazing it is

04:09.200 --> 04:11.520
 to be able to have the conversations

04:11.520 --> 04:14.640
 that I have with the people I get to have them with.

04:14.640 --> 04:17.040
 You had to reach back into the sentimental

04:17.040 --> 04:21.640
 and what's the radical markets and the economics?

04:21.640 --> 04:24.760
 So the idea with radical markets is like,

04:24.760 --> 04:29.760
 can you come up with new market based mechanisms to,

04:32.320 --> 04:33.800
 you know, I think we have this,

04:33.800 --> 04:35.240
 we're having this debate right now,

04:35.240 --> 04:40.040
 like does capitalism work, like free markets work?

04:40.040 --> 04:43.000
 Can the incentive structures

04:43.000 --> 04:46.360
 that are built into these systems produce outcomes

04:46.360 --> 04:51.360
 that are creating sort of equitably distributed benefits

04:51.560 --> 04:53.520
 for every member of society?

04:55.400 --> 04:58.720
 You know, and I think it's a reasonable set of questions

04:58.720 --> 04:59.560
 to be asking.

04:59.560 --> 05:02.160
 And so what Glenn, and so like, you know,

05:02.160 --> 05:04.400
 one mode of thought there, like if you have doubts

05:04.400 --> 05:06.720
 that the markets are actually working,

05:06.720 --> 05:08.560
 you can sort of like tip towards like,

05:08.560 --> 05:10.800
 okay, let's become more socialist

05:10.800 --> 05:14.240
 and like have central planning and governments

05:14.240 --> 05:15.800
 or some other central organization

05:15.800 --> 05:18.280
 is like making a bunch of decisions

05:18.280 --> 05:22.040
 about how sort of work gets done

05:22.040 --> 05:25.400
 and like where the investments

05:25.400 --> 05:28.880
 and where the outputs of those investments get distributed.

05:28.880 --> 05:32.160
 Glenn's notion is like lean more

05:32.160 --> 05:35.800
 into like the market based mechanism.

05:35.800 --> 05:37.920
 So like for instance,

05:37.920 --> 05:39.600
 this is one of the more radical ideas,

05:39.600 --> 05:44.600
 like suppose that you had a radical pricing mechanism

05:45.160 --> 05:47.120
 for assets like real estate

05:47.120 --> 05:52.120
 where you could be bid out of your position

05:53.600 --> 05:58.600
 in your home, you know, for instance.

05:58.720 --> 06:01.120
 So like if somebody came along and said,

06:01.120 --> 06:04.400
 you know, like I can find higher economic utility

06:04.400 --> 06:05.760
 for this piece of real estate

06:05.760 --> 06:08.720
 that you're running your business in,

06:08.720 --> 06:13.040
 like then like you either have to, you know,

06:13.040 --> 06:16.440
 sort of bid to sort of stay

06:16.440 --> 06:19.960
 or like the thing that's got the higher economic utility,

06:19.960 --> 06:21.440
 you know, sort of takes over the asset

06:21.440 --> 06:23.720
 and which would make it very difficult

06:23.720 --> 06:27.600
 to have the same sort of rent seeking behaviors

06:27.600 --> 06:29.000
 that you've got right now

06:29.000 --> 06:34.000
 because like if you did speculative bidding,

06:34.000 --> 06:39.000
 like you would very quickly like lose a whole lot of money.

06:40.440 --> 06:43.520
 And so like the prices of the assets would be sort of

06:43.520 --> 06:47.600
 like very closely indexed to like the value

06:47.600 --> 06:49.720
 that they can produce.

06:49.720 --> 06:52.680
 And like because like you'd have this sort of real time

06:52.680 --> 06:55.320
 mechanism that would force you to sort of mark the value

06:55.320 --> 06:56.800
 of the asset to the market,

06:56.800 --> 06:58.560
 then it could be taxed appropriately.

06:58.560 --> 07:00.400
 Like you couldn't sort of sit on this thing and say,

07:00.400 --> 07:03.040
 oh, like this house is only worth 10,000 bucks

07:03.040 --> 07:06.600
 when like everything around it is worth 10 million.

07:06.600 --> 07:07.440
 That's really interesting.

07:07.440 --> 07:08.720
 So it's an incentive structure

07:08.720 --> 07:13.200
 that where the prices match the value much better.

07:13.200 --> 07:14.040
 Yeah.

07:14.040 --> 07:16.320
 And Glenn does a much, much better job than I do

07:16.320 --> 07:18.920
 at selling and I probably picked the world's worst example,

07:18.920 --> 07:20.360
 you know, and, and, and, but like,

07:20.360 --> 07:24.520
 and it's intentionally provocative, you know,

07:24.520 --> 07:26.480
 so like this whole notion, like I, you know,

07:26.480 --> 07:28.920
 like I'm not sure whether I like this notion

07:28.920 --> 07:31.120
 that like we can have a set of market mechanisms

07:31.120 --> 07:35.360
 where I could get bid out of, out of my property, you know,

07:35.360 --> 07:37.680
 but, but, you know, like if you're thinking about something

07:37.680 --> 07:42.480
 like Elizabeth Warren's wealth tax, for instance,

07:42.480 --> 07:45.600
 like you would have, I mean, it'd be really interesting

07:45.600 --> 07:50.080
 in like how you would actually set the price on the assets.

07:50.080 --> 07:52.040
 And like you might have to have a mechanism like that

07:52.040 --> 07:54.160
 if you put a tax like that in place.

07:54.160 --> 07:56.440
 It's really interesting that that kind of research,

07:56.440 --> 07:59.800
 at least tangentially touching Microsoft research.

07:59.800 --> 08:00.640
 Yeah.

08:00.640 --> 08:02.560
 So if you're really thinking broadly,

08:02.560 --> 08:07.560
 maybe you can speak to this connects to AI.

08:08.400 --> 08:10.680
 So we have a candidate, Andrew Yang,

08:10.680 --> 08:13.480
 who kind of talks about artificial intelligence

08:13.480 --> 08:16.640
 and the concern that people have about, you know,

08:16.640 --> 08:19.000
 automations impact on society.

08:19.000 --> 08:22.680
 And arguably Microsoft is at the cutting edge

08:22.680 --> 08:25.040
 of innovation in all these kinds of ways.

08:25.040 --> 08:27.080
 And so it's pushing AI forward.

08:27.080 --> 08:30.040
 How do you think about combining all our conversations

08:30.040 --> 08:32.840
 together here with radical markets and socialism

08:32.840 --> 08:37.520
 and innovation in AI that Microsoft is doing?

08:37.520 --> 08:42.520
 And then Andrew Yang's worry that that will,

08:43.520 --> 08:46.840
 that will result in job loss for the lower and so on.

08:46.840 --> 08:47.680
 How do you think about that?

08:47.680 --> 08:51.160
 I think it's sort of one of the most important questions

08:51.160 --> 08:55.320
 in technology, like maybe even in society right now

08:55.320 --> 09:00.320
 about how is AI going to develop over the course

09:00.720 --> 09:02.000
 of the next several decades

09:02.000 --> 09:03.600
 and like what's it gonna be used for

09:03.600 --> 09:06.560
 and like what benefits will it produce

09:06.560 --> 09:08.520
 and what negative impacts will it produce

09:08.520 --> 09:13.520
 and you know, who gets to steer this whole thing?

09:13.720 --> 09:16.320
 You know, I'll say at the highest level,

09:17.240 --> 09:22.240
 one of the real joys of getting to do what I do at Microsoft

09:22.240 --> 09:27.240
 is Microsoft has this heritage as a platform company.

09:27.560 --> 09:31.040
 And so, you know, like Bill has this thing

09:31.040 --> 09:32.880
 that he said a bunch of years ago

09:32.880 --> 09:36.440
 where the measure of a successful platform

09:36.440 --> 09:39.800
 is that it produces far more economic value

09:39.800 --> 09:41.840
 for the people who build on top of the platform

09:41.840 --> 09:46.840
 than is created for the platform owner or builder.

09:47.320 --> 09:50.920
 And I think we have to think about AI that way.

09:50.920 --> 09:55.920
 Like it has to be a platform that other people can use

09:56.280 --> 10:01.280
 to build businesses, to fulfill their creative objectives,

10:01.280 --> 10:04.640
 to be entrepreneurs, to solve problems that they have

10:04.640 --> 10:07.680
 in their work and in their lives.

10:07.680 --> 10:11.960
 It can't be a thing where there are a handful of companies

10:11.960 --> 10:16.440
 sitting in a very small handful of cities geographically

10:16.440 --> 10:19.120
 who are making all the decisions

10:19.120 --> 10:24.120
 about what goes into the AI and like,

10:24.240 --> 10:26.920
 and then on top of like all this infrastructure,

10:26.920 --> 10:31.000
 then build all of the commercially valuable uses for it.

10:31.000 --> 10:34.400
 So like, I think like that's bad from a, you know,

10:34.400 --> 10:36.520
 sort of, you know, economics

10:36.520 --> 10:39.720
 and sort of equitable distribution of value perspective,

10:39.720 --> 10:42.080
 like, you know, sort of back to this whole notion of,

10:42.080 --> 10:44.560
 you know, like, do the markets work?

10:44.560 --> 10:47.600
 But I think it's also bad from an innovation perspective

10:47.600 --> 10:51.360
 because like I have infinite amounts of faith

10:51.360 --> 10:53.880
 in human beings that if you, you know,

10:53.880 --> 10:58.280
 give folks powerful tools, they will go do interesting things.

10:58.280 --> 11:02.320
 And it's more than just a few tens of thousands of people

11:02.320 --> 11:03.360
 with the interesting tools,

11:03.360 --> 11:05.400
 it should be millions of people with the tools.

11:05.400 --> 11:07.200
 So it's sort of like, you know,

11:07.200 --> 11:10.200
 you think about the steam engine

11:10.200 --> 11:13.800
 and the late 18th century, like it was, you know,

11:13.800 --> 11:16.800
 maybe the first large scale substitute for human labor

11:16.800 --> 11:19.120
 that we've built like a machine.

11:19.120 --> 11:21.680
 And, you know, in the beginning,

11:21.680 --> 11:23.520
 when these things are getting deployed,

11:23.520 --> 11:28.320
 the folks who got most of the value from the steam engines

11:28.320 --> 11:30.160
 were the folks who had capital

11:30.160 --> 11:31.600
 so they could afford to build them.

11:31.600 --> 11:34.720
 And like they built factories around them in businesses

11:34.720 --> 11:38.680
 and the experts who knew how to build and maintain them.

11:38.680 --> 11:42.880
 But access to that technology democratized over time.

11:42.880 --> 11:47.040
 Like now like an engine is not a,

11:47.040 --> 11:48.800
 it's not like a differentiated thing.

11:48.800 --> 11:50.280
 Like there isn't one engine company

11:50.280 --> 11:51.560
 that builds all the engines

11:51.560 --> 11:53.120
 and all of the things that use engines

11:53.120 --> 11:54.240
 are made by this company.

11:54.240 --> 11:57.440
 And like they get all the economics from all of that.

11:57.440 --> 11:59.320
 Like, no, like fully demarcated.

11:59.320 --> 12:00.600
 Like they're probably, you know,

12:00.600 --> 12:02.360
 we're sitting here in this room

12:02.360 --> 12:03.680
 and like even though they don't,

12:03.680 --> 12:05.280
 they're probably things, you know,

12:05.280 --> 12:09.120
 like the MIMS gyroscope that are in both of our,

12:09.120 --> 12:11.480
 like there's like little engines, you know,

12:11.480 --> 12:14.520
 sort of everywhere, they're just a component

12:14.520 --> 12:16.240
 in how we build the modern world.

12:16.240 --> 12:17.680
 Like AI needs to get there.

12:17.680 --> 12:20.200
 Yeah, so that's a really powerful way to think.

12:20.200 --> 12:25.120
 If we think of AI as a platform versus a tool

12:25.120 --> 12:27.600
 that Microsoft owns as a platform

12:27.600 --> 12:30.120
 that enables creation on top of it,

12:30.120 --> 12:31.520
 that's the way to democratize it.

12:31.520 --> 12:34.200
 That's really interesting actually.

12:34.200 --> 12:36.040
 And Microsoft throughout its history

12:36.040 --> 12:38.240
 has been positioned well to do that.

12:38.240 --> 12:41.640
 And the, you know, the tieback to this radical markets thing,

12:41.640 --> 12:46.640
 like the, so my team has been working with Glenn

12:47.800 --> 12:51.120
 on this and Jaren Lanier actually.

12:51.120 --> 12:56.120
 So Jaren is the like the sort of father of virtual reality.

12:56.440 --> 12:59.480
 Like he's one of the most interesting human beings

12:59.480 --> 13:01.760
 on the planet, like a sweet, sweet guy.

13:02.840 --> 13:07.120
 And so Jaren and Glenn and folks in my team

13:07.120 --> 13:10.360
 have been working on this notion of data as labor

13:10.360 --> 13:13.160
 or like they call it data dignity as well.

13:13.160 --> 13:16.880
 And so the idea is that if you, you know,

13:16.880 --> 13:18.600
 again, going back to this, you know,

13:18.600 --> 13:20.800
 sort of industrial analogy,

13:20.800 --> 13:23.560
 if you think about data as the raw material

13:23.560 --> 13:27.640
 that is consumed by the machine of AI

13:27.640 --> 13:30.560
 in order to do useful things,

13:30.560 --> 13:34.400
 then like we're not doing a really great job right now

13:34.400 --> 13:37.760
 in having transparent marketplaces for valuing

13:37.760 --> 13:39.800
 those data contributions.

13:39.800 --> 13:42.680
 So like, and we all make them like explicitly,

13:42.680 --> 13:43.600
 like you go to LinkedIn,

13:43.600 --> 13:46.160
 you sort of set up your profile on LinkedIn,

13:46.160 --> 13:47.800
 like that's an explicit contribution.

13:47.800 --> 13:49.480
 Like, you know exactly the information

13:49.480 --> 13:50.720
 that you're putting into the system.

13:50.720 --> 13:53.000
 And like you put it there because you have

13:53.000 --> 13:55.520
 some nominal notion of like what value

13:55.520 --> 13:56.640
 you're going to get in return,

13:56.640 --> 13:57.720
 but it's like only nominal.

13:57.720 --> 13:59.680
 Like you don't know exactly what value

13:59.680 --> 14:02.040
 you're getting in return, like services free, you know,

14:02.040 --> 14:04.600
 like it's low amount of like perceived.

14:04.600 --> 14:06.680
 And then you've got all this indirect contribution

14:06.680 --> 14:08.960
 that you're making just by virtue of interacting

14:08.960 --> 14:13.160
 with all of the technology that's in your daily life.

14:13.160 --> 14:16.120
 And so like what Glenn and Jaren

14:16.120 --> 14:19.440
 and this data dignity team are trying to do is like,

14:19.440 --> 14:22.240
 can we figure out a set of mechanisms

14:22.240 --> 14:26.000
 that let us value those data contributions

14:26.000 --> 14:28.200
 so that you could create an economy

14:28.200 --> 14:31.480
 and like a set of controls and incentives

14:31.480 --> 14:36.480
 that would allow people to like maybe even in the limit

14:36.840 --> 14:38.880
 like earn part of their living

14:38.880 --> 14:41.000
 through the data that they're creating.

14:41.000 --> 14:42.680
 And like you can sort of see it in explicit ways.

14:42.680 --> 14:46.000
 There are these companies like Scale AI

14:46.000 --> 14:49.960
 and like they're a whole bunch of them in China right now

14:49.960 --> 14:52.400
 that are basically data labeling companies.

14:52.400 --> 14:54.560
 So like you're doing supervised machine learning,

14:54.560 --> 14:57.400
 you need lots and lots of label training data.

14:58.600 --> 15:01.440
 And like those people are getting like who work

15:01.440 --> 15:03.600
 for those companies are getting compensated

15:03.600 --> 15:06.360
 for their data contributions into the system.

15:06.360 --> 15:07.720
 And so...

15:07.720 --> 15:10.280
 That's easier to put a number on their contribution

15:10.280 --> 15:11.960
 because they're explicitly labeling data.

15:11.960 --> 15:12.800
 Correct.

15:12.800 --> 15:14.360
 But you're saying that we're all contributing data

15:14.360 --> 15:15.720
 in different kinds of ways.

15:15.720 --> 15:19.640
 And it's fascinating to start to explicitly try

15:19.640 --> 15:20.880
 to put a number on it.

15:20.880 --> 15:22.600
 Do you think that's possible?

15:22.600 --> 15:23.640
 I don't know, it's hard.

15:23.640 --> 15:25.480
 It really is.

15:25.480 --> 15:30.480
 Because, you know, we don't have as much transparency

15:30.480 --> 15:35.480
 as I think we need in like how the data is getting used.

15:37.240 --> 15:38.720
 And it's, you know, super complicated.

15:38.720 --> 15:41.000
 Like, you know, we, you know,

15:41.000 --> 15:42.880
 I think as technologists sort of appreciate

15:42.880 --> 15:44.160
 like some of the subtlety there.

15:44.160 --> 15:47.880
 It's like, you know, the data, the data gets created

15:47.880 --> 15:51.400
 and then it gets, you know, it's not valuable.

15:51.400 --> 15:56.000
 Like the data exhaust that you give off

15:56.000 --> 15:58.480
 or the, you know, the explicit data

15:58.480 --> 16:03.240
 that I am putting into the system isn't valuable.

16:03.240 --> 16:05.160
 It's super valuable atomically.

16:05.160 --> 16:08.360
 Like it's only valuable when you sort of aggregate it together

16:08.360 --> 16:10.440
 into, you know, sort of large numbers.

16:10.440 --> 16:11.960
 It's true even for these like folks

16:11.960 --> 16:14.880
 who are getting compensated for like labeling things.

16:14.880 --> 16:16.480
 Like for supervised machine learning now,

16:16.480 --> 16:20.080
 like you need lots of labels to train, you know,

16:20.080 --> 16:22.080
 a model that performs well.

16:22.080 --> 16:24.440
 And so, you know, I think that's one of the challenges.

16:24.440 --> 16:26.120
 It's like, how do you, you know,

16:26.120 --> 16:28.000
 how do you sort of figure out like

16:28.000 --> 16:31.480
 because this data is getting combined in so many ways,

16:31.480 --> 16:33.880
 like through these combinations,

16:33.880 --> 16:35.880
 like how the value is flowing.

16:35.880 --> 16:38.520
 Yeah, that's, that's fascinating.

16:38.520 --> 16:39.360
 Yeah.

16:39.360 --> 16:41.880
 And it's fascinating that you're thinking about this.

16:41.880 --> 16:44.160
 And I wasn't even going into this competition

16:44.160 --> 16:48.200
 expecting the breadth of research really

16:48.200 --> 16:50.600
 that Microsoft broadly is thinking about.

16:50.600 --> 16:52.360
 You are thinking about in Microsoft.

16:52.360 --> 16:57.360
 So if we go back to 89 when Microsoft released Office

16:57.360 --> 17:00.920
 or 1990 when they released Windows 3.0,

17:00.920 --> 17:04.960
 how's the, in your view,

17:04.960 --> 17:07.280
 I know you weren't there the entire, you know,

17:07.280 --> 17:09.760
 through its history, but how has the company changed

17:09.760 --> 17:12.840
 in the 30 years since as you look at it now?

17:12.840 --> 17:17.080
 The good thing is it's started off as a platform company.

17:17.080 --> 17:19.960
 Like it's still a platform company,

17:19.960 --> 17:22.640
 like the parts of the business that are like thriving

17:22.640 --> 17:26.560
 and most successful or those that are building platforms,

17:26.560 --> 17:29.000
 like the mission of the company now is,

17:29.000 --> 17:30.120
 the mission's changed.

17:30.120 --> 17:32.480
 It's like changing a very interesting way.

17:32.480 --> 17:36.280
 So, you know, back in 89.90,

17:36.280 --> 17:39.040
 like they were still on the original mission,

17:39.040 --> 17:43.840
 which was like put a PC on every desk and in every home.

17:43.840 --> 17:47.480
 Like, and it was basically about democratizing access

17:47.480 --> 17:50.000
 to this new personal computing technology,

17:50.000 --> 17:52.680
 which when Bill started the company,

17:52.680 --> 17:57.680
 integrated circuit microprocessors were a brand new thing

17:57.680 --> 18:00.120
 and like people were building, you know,

18:00.120 --> 18:03.840
 homebrew computers, you know, from kits,

18:03.840 --> 18:07.520
 like the way people build ham radios right now.

18:08.520 --> 18:10.680
 And I think this is sort of the interesting thing

18:10.680 --> 18:12.840
 for folks who build platforms in general.

18:12.840 --> 18:16.840
 Bill saw the opportunity there

18:16.840 --> 18:18.720
 and what personal computers could do.

18:18.720 --> 18:20.440
 And it was like, it was sort of a reach.

18:20.440 --> 18:21.680
 Like you just sort of imagined

18:21.680 --> 18:23.880
 like where things were, you know,

18:23.880 --> 18:24.880
 when they started the company

18:24.880 --> 18:26.120
 versus where things are now.

18:26.120 --> 18:29.400
 Like in success, when you democratize a platform,

18:29.400 --> 18:31.000
 it just sort of vanishes into the platform.

18:31.000 --> 18:32.480
 You don't pay attention to it anymore.

18:32.480 --> 18:35.600
 Like operating systems aren't a thing anymore.

18:35.600 --> 18:38.040
 Like they're super important, like completely critical.

18:38.040 --> 18:41.760
 And like, you know, when you see one, you know, fail,

18:41.760 --> 18:43.520
 like you just, you sort of understand,

18:43.520 --> 18:45.320
 but like, you know, it's not a thing where you're,

18:45.320 --> 18:47.920
 you're not like waiting for, you know,

18:47.920 --> 18:50.480
 the next operating system thing

18:50.480 --> 18:52.960
 in the same way that you were in 1995, right?

18:52.960 --> 18:54.280
 Like in 1995, like, you know,

18:54.280 --> 18:56.000
 we had Rolling Stones on the stage

18:56.000 --> 18:57.600
 with the Windows 95 roll out.

18:57.600 --> 18:59.320
 Like it was like the biggest thing in the world.

18:59.320 --> 19:01.080
 Everybody would like lined up for it

19:01.080 --> 19:03.400
 the way that people used to line up for iPhone.

19:03.400 --> 19:05.120
 But like, you know, eventually,

19:05.120 --> 19:07.160
 and like this isn't necessarily a bad thing.

19:07.160 --> 19:09.000
 Like it just sort of, you know,

19:09.000 --> 19:12.880
 the success is that it's sort of, it becomes ubiquitous.

19:12.880 --> 19:14.800
 It's like everywhere and like human beings

19:14.800 --> 19:16.640
 when their technology becomes ubiquitous,

19:16.640 --> 19:18.240
 they just sort of start taking it for granted.

19:18.240 --> 19:23.240
 So the mission now that Satya rearticulated

19:23.640 --> 19:25.280
 five plus years ago now

19:25.280 --> 19:27.360
 when he took over as CEO of the company,

19:29.320 --> 19:33.480
 our mission is to empower every individual

19:33.480 --> 19:37.760
 and every organization in the world to be more successful.

19:39.200 --> 19:43.160
 And so, you know, again, like that's a platform mission.

19:43.160 --> 19:46.320
 And like the way that we do it now is different.

19:46.320 --> 19:48.680
 It's like we have a hyperscale cloud

19:48.680 --> 19:51.680
 that people are building their applications on top of.

19:51.680 --> 19:53.680
 Like we have a bunch of AI infrastructure

19:53.680 --> 19:56.280
 that people are building their AI applications on top of.

19:56.280 --> 20:01.280
 We have, you know, we have a productivity suite of software

20:02.280 --> 20:05.800
 like Microsoft Dynamics, which, you know,

20:05.800 --> 20:07.440
 some people might not think is the sexiest thing

20:07.440 --> 20:10.040
 in the world, but it's like helping people figure out

20:10.040 --> 20:12.720
 how to automate all of their business processes

20:12.720 --> 20:16.800
 and workflows and to, you know, like help those businesses

20:16.800 --> 20:19.120
 using it to like grow and be more successful.

20:19.120 --> 20:24.120
 So it's a much broader vision in a way now

20:24.240 --> 20:25.480
 than it was back then.

20:25.480 --> 20:27.400
 Like it was sort of very particular thing.

20:27.400 --> 20:29.280
 And like now, like we live in this world

20:29.280 --> 20:31.320
 where technology is so powerful

20:31.320 --> 20:36.320
 and it's like such a basic fact of life

20:36.320 --> 20:39.760
 that it, you know, that it both exists

20:39.760 --> 20:42.760
 and is going to get better and better over time

20:42.760 --> 20:46.000
 or at least more and more powerful over time.

20:46.000 --> 20:48.200
 So like, you know, what you have to do as a platform player

20:48.200 --> 20:49.920
 is just much bigger.

20:49.920 --> 20:50.760
 Right.

20:50.760 --> 20:52.600
 There's so many directions in which you can transform.

20:52.600 --> 20:55.160
 You didn't mention mixed reality too.

20:55.160 --> 20:59.200
 You know, that's probably early days

20:59.200 --> 21:00.680
 or depends how you think of it.

21:00.680 --> 21:02.240
 But if we think in a scale of centuries,

21:02.240 --> 21:04.120
 it's the early days of mixed reality.

21:04.120 --> 21:04.960
 Oh, for sure.

21:04.960 --> 21:08.280
 And so yeah, with how it lands,

21:08.280 --> 21:10.600
 the Microsoft is doing some really interesting work there.

21:10.600 --> 21:13.560
 Do you touch that part of the effort?

21:13.560 --> 21:14.840
 What's the thinking?

21:14.840 --> 21:17.640
 Do you think of mixed reality as a platform too?

21:17.640 --> 21:18.480
 Oh, sure.

21:18.480 --> 21:21.320
 When we look at what the platforms of the future could be.

21:21.320 --> 21:23.880
 So like fairly obvious that like AI is one,

21:23.880 --> 21:26.600
 like you don't have to, I mean, like that's,

21:26.600 --> 21:29.160
 you know, you sort of say it to like someone

21:29.160 --> 21:31.920
 and you know, like they get it.

21:31.920 --> 21:36.280
 But like we also think of the like mixed reality

21:36.280 --> 21:39.560
 and quantum is like these two interesting,

21:39.560 --> 21:40.920
 you know, potentially.

21:40.920 --> 21:41.800
 Quantum computing.

21:41.800 --> 21:42.640
 Yeah.

21:42.640 --> 21:44.520
 Okay, so let's get crazy then.

21:44.520 --> 21:48.920
 So you're talking about some futuristic things here.

21:48.920 --> 21:50.920
 Well, the mixed reality Microsoft is really,

21:50.920 --> 21:52.600
 it's not even futuristic, it's here.

21:52.600 --> 21:53.440
 It is.

21:53.440 --> 21:54.280
 Incredible stuff.

21:54.280 --> 21:56.680
 And look, and it's having an impact right now.

21:56.680 --> 21:58.720
 Like one of the more interesting things

21:58.720 --> 22:01.280
 that's happened with mixed reality over the past

22:01.280 --> 22:04.120
 couple of years that I didn't clearly see

22:04.120 --> 22:08.400
 is that it's become the computing device

22:08.400 --> 22:13.160
 for folks who, for doing their work

22:13.160 --> 22:16.040
 who haven't used any computing device at all

22:16.040 --> 22:16.960
 to do their work before.

22:16.960 --> 22:19.800
 So technicians and service folks

22:19.800 --> 22:24.200
 and people who are doing like machine maintenance

22:24.200 --> 22:25.280
 on factory floors.

22:25.280 --> 22:28.760
 So like they, you know, because they're mobile

22:28.760 --> 22:30.280
 and like they're out in the world

22:30.280 --> 22:32.320
 and they're working with their hands

22:32.320 --> 22:34.080
 and, you know, sort of servicing these

22:34.080 --> 22:36.520
 like very complicated things.

22:36.520 --> 22:39.440
 They're, they don't use their mobile phone

22:39.440 --> 22:41.440
 and like they don't carry a laptop with them.

22:41.440 --> 22:43.480
 And, you know, they're not tethered to a desk.

22:43.480 --> 22:46.920
 And so mixed reality, like where it's getting

22:46.920 --> 22:48.840
 traction right now, where HoloLens is selling

22:48.840 --> 22:53.840
 a lot of units is for these sorts of applications

22:53.880 --> 22:55.440
 for these workers and it's become like,

22:55.440 --> 22:58.040
 I mean, like the people love it.

22:58.040 --> 23:00.600
 They're like, oh my God, like this is like,

23:00.600 --> 23:02.840
 for them like the same sort of productivity boosts

23:02.840 --> 23:05.520
 that, you know, like an office worker had

23:05.520 --> 23:08.200
 when they got their first personal computer.

23:08.200 --> 23:09.800
 Yeah, but you did mention,

23:09.800 --> 23:13.400
 it's certainly obvious AI as a platform,

23:13.400 --> 23:15.560
 but can we dig into it a little bit?

23:15.560 --> 23:18.320
 How does AI begin to infuse some of the products

23:18.320 --> 23:19.480
 in Microsoft?

23:19.480 --> 23:24.480
 So currently providing training of, for example,

23:25.040 --> 23:26.760
 neural networks in the cloud

23:26.760 --> 23:30.960
 or providing pre trained models

23:30.960 --> 23:35.360
 or just even providing computing resources

23:35.360 --> 23:37.520
 and whatever different inference

23:37.520 --> 23:39.320
 that you want to do using neural networks.

23:39.320 --> 23:40.160
 Yep.

23:40.160 --> 23:43.560
 Well, how do you think of AI infusing the,

23:43.560 --> 23:45.880
 as a platform that Microsoft can provide?

23:45.880 --> 23:48.320
 Yeah, I mean, I think it's, it's super interesting.

23:48.320 --> 23:49.560
 It's like everywhere.

23:49.560 --> 23:54.560
 And like we run these, we run these review meetings now

23:54.560 --> 23:59.560
 where it's me and Satya and like members of Satya's

24:01.480 --> 24:04.600
 leadership team and like a cross functional group

24:04.600 --> 24:06.200
 of folks across the entire company

24:06.200 --> 24:11.200
 who are working on like either AI infrastructure

24:11.840 --> 24:15.520
 or like have some substantial part of their,

24:16.480 --> 24:21.480
 of their product work using AI in some significant way.

24:21.480 --> 24:23.440
 Now, the important thing to understand is like,

24:23.440 --> 24:27.040
 when you think about like how the AI is going to manifest

24:27.040 --> 24:29.600
 in like an experience for something

24:29.600 --> 24:30.760
 that's going to make it better,

24:30.760 --> 24:35.760
 like I think you don't want the AI in this

24:35.760 --> 24:37.760
 to be the first order thing.

24:37.760 --> 24:40.600
 It's like whatever the product is and like the thing

24:40.600 --> 24:42.440
 that is trying to help you do,

24:42.440 --> 24:44.560
 like the AI just sort of makes it better.

24:44.560 --> 24:46.840
 And you know, this is a gross exaggeration,

24:46.840 --> 24:50.680
 but like I, yeah, people get super excited about it.

24:50.680 --> 24:53.280
 They're super excited about like where the AI is showing up

24:53.280 --> 24:55.440
 in products and I'm like, do you get that excited

24:55.440 --> 24:59.880
 about like where you're using a hash table like in your code?

24:59.880 --> 25:03.200
 Like it's just another, it's a very interesting

25:03.200 --> 25:05.800
 programming tool, but it's sort of like it's an engineering

25:05.800 --> 25:09.560
 tool and so like it shows up everywhere.

25:09.560 --> 25:12.920
 So like we've got dozens and dozens of features now

25:12.920 --> 25:17.400
 in office that are powered by like fairly sophisticated

25:17.400 --> 25:22.200
 machine learning, our search engine wouldn't work at all

25:22.200 --> 25:24.840
 if you took the machine learning out of it.

25:24.840 --> 25:28.560
 The like increasingly, you know,

25:28.560 --> 25:33.560
 things like content moderation on our Xbox and xCloud

25:34.800 --> 25:35.960
 platform.

25:37.000 --> 25:39.160
 When you mean moderation to me, like the recommender

25:39.160 --> 25:41.760
 is like showing what you want to look at next.

25:41.760 --> 25:44.000
 No, no, no, it's like anti bullying stuff.

25:44.000 --> 25:47.040
 So the usual social network stuff that you have to deal with.

25:47.040 --> 25:47.880
 Yeah, correct.

25:47.880 --> 25:50.080
 But it's like really it's targeted,

25:50.080 --> 25:52.280
 it's targeted towards a gaming audience.

25:52.280 --> 25:55.320
 So it's like a very particular type of thing where,

25:55.320 --> 25:59.480
 you know, the the line between playful banter

25:59.480 --> 26:02.280
 and like legitimate bullying is like a subtle one.

26:02.280 --> 26:06.080
 And like you have to, it's sort of tough.

26:06.080 --> 26:09.080
 Like I have, I love to, if we could dig into it

26:09.080 --> 26:11.720
 because you're also, you led the engineering efforts

26:11.720 --> 26:14.920
 of LinkedIn and if we look at,

26:14.920 --> 26:17.640
 if we look at LinkedIn as a social network

26:17.640 --> 26:21.760
 and if we look at the Xbox gaming as the social components,

26:21.760 --> 26:24.840
 the very different kinds of, I imagine communication

26:24.840 --> 26:26.880
 going on on the two platforms, right?

26:26.880 --> 26:29.520
 And the line in terms of bullying and so on

26:29.520 --> 26:31.480
 is different on the two platforms.

26:31.480 --> 26:33.480
 So how do you, I mean,

26:33.480 --> 26:36.240
 such a fascinating philosophical discussion

26:36.240 --> 26:37.240
 of where that line is.

26:37.240 --> 26:39.840
 I don't think anyone knows the right answer.

26:39.840 --> 26:42.040
 Twitter folks are under fire now,

26:42.040 --> 26:45.120
 Jack at Twitter for trying to find that line.

26:45.120 --> 26:46.920
 Nobody knows what that line is,

26:46.920 --> 26:51.720
 but how do you try to find the line for,

26:52.480 --> 26:57.480
 you know, trying to prevent abusive behavior

26:58.040 --> 27:00.200
 and at the same time let people be playful

27:00.200 --> 27:02.880
 and joke around and that kind of thing.

27:02.880 --> 27:04.640
 I think in a certain way, like, you know,

27:04.640 --> 27:09.640
 if you have what I would call vertical social networks,

27:09.640 --> 27:12.200
 it gets to be a little bit easier.

27:12.200 --> 27:14.440
 So like if you have a clear notion

27:14.440 --> 27:17.960
 of like what your social network should be used for

27:17.960 --> 27:22.280
 or like what you are designing a community around,

27:22.280 --> 27:25.800
 then you don't have as many dimensions

27:25.800 --> 27:28.960
 to your sort of content safety problem

27:28.960 --> 27:33.720
 as, you know, as you do in a general purpose platform.

27:33.720 --> 27:37.520
 I mean, so like on LinkedIn,

27:37.520 --> 27:39.920
 like the whole social network is about

27:39.920 --> 27:41.560
 connecting people with opportunity,

27:41.560 --> 27:43.160
 whether it's helping them find a job

27:43.160 --> 27:46.280
 or to, you know, sort of find mentors

27:46.280 --> 27:49.320
 or to, you know, sort of help them

27:49.320 --> 27:52.120
 like find their next sales lead

27:52.120 --> 27:56.160
 or to just sort of allow them to broadcast

27:56.160 --> 27:59.440
 their, you know, sort of professional identity

27:59.440 --> 28:04.440
 to their network of peers and collaborators

28:04.440 --> 28:05.880
 and, you know, sort of professional community.

28:05.880 --> 28:07.400
 Like that is, I mean, like in some ways,

28:07.400 --> 28:08.960
 like that's very, very broad,

28:08.960 --> 28:12.480
 but in other ways, it's sort of, you know, it's narrow.

28:12.480 --> 28:17.480
 And so like you can build AIs like machine learning systems

28:18.360 --> 28:23.360
 that are, you know, capable with those boundaries

28:23.360 --> 28:26.200
 of making better automated decisions about like,

28:26.200 --> 28:28.240
 what is, you know, sort of inappropriate

28:28.240 --> 28:30.440
 and offensive comment or dangerous comment

28:30.440 --> 28:31.920
 or illegal content.

28:31.920 --> 28:34.800
 When you have some constraints,

28:34.800 --> 28:37.400
 you know, same thing with, you know,

28:37.400 --> 28:40.880
 same thing with like the gaming social network.

28:40.880 --> 28:42.680
 So for instance, like it's about playing games,

28:42.680 --> 28:44.880
 about having fun and like the thing

28:44.880 --> 28:47.240
 that you don't want to have happen on the platform.

28:47.240 --> 28:49.160
 It's why bullying is such an important thing.

28:49.160 --> 28:50.600
 Like bullying is not fun.

28:50.600 --> 28:53.400
 So you want to do everything in your power

28:53.400 --> 28:56.240
 to encourage that not to happen.

28:56.240 --> 29:00.320
 And yeah, but I think that's a really important thing

29:00.320 --> 29:03.920
 but I think it's sort of a tough problem in general.

29:03.920 --> 29:05.280
 It's one where I think, you know,

29:05.280 --> 29:07.120
 eventually we're gonna have to have

29:09.120 --> 29:13.800
 some sort of clarification from our policy makers

29:13.800 --> 29:17.400
 about what it is that we should be doing,

29:17.400 --> 29:20.880
 like where the lines are, because it's tough.

29:20.880 --> 29:23.760
 Like you don't, like in democracy, right?

29:23.760 --> 29:26.680
 Like you don't want, you want some sort

29:26.680 --> 29:28.880
 of democratic involvement.

29:28.880 --> 29:30.440
 Like people should have a say

29:30.440 --> 29:34.680
 in like where the lines are drawn.

29:34.680 --> 29:36.920
 Like you don't want a bunch of people

29:36.920 --> 29:39.480
 making like unilateral decisions.

29:39.480 --> 29:43.120
 And like we are in a state right now

29:43.120 --> 29:44.760
 for some of these platforms where you actually

29:44.760 --> 29:46.280
 do have to make unilateral decisions

29:46.280 --> 29:48.640
 where the policy making isn't gonna happen fast enough

29:48.640 --> 29:52.520
 in order to like prevent very bad things from happening.

29:52.520 --> 29:55.200
 But like we need the policy making side of that

29:55.200 --> 29:58.480
 to catch up I think as quickly as possible

29:58.480 --> 30:00.680
 because you want that whole process

30:00.680 --> 30:02.000
 to be a democratic thing,

30:02.000 --> 30:05.760
 not a, you know, not some sort of weird thing

30:05.760 --> 30:08.040
 where you've got a non representative group

30:08.040 --> 30:10.440
 of people making decisions that have, you know,

30:10.440 --> 30:12.520
 like national and global impact.

30:12.520 --> 30:14.720
 And it's fascinating because the digital space

30:14.720 --> 30:17.520
 is different than the physical space

30:17.520 --> 30:19.800
 in which nations and governments were established.

30:19.800 --> 30:23.960
 And so what policy looks like globally,

30:23.960 --> 30:25.760
 what bullying looks like globally,

30:25.760 --> 30:28.360
 what healthy communication looks like globally

30:28.360 --> 30:31.920
 is an open question and we're all figuring it out together.

30:31.920 --> 30:32.760
 Which is fascinating.

30:32.760 --> 30:37.160
 Yeah, I mean with, you know, sort of fake news for instance

30:37.160 --> 30:42.160
 and deep fakes and fake news generated by humans.

30:42.320 --> 30:44.600
 Yeah, so we can talk about deep fakes.

30:44.600 --> 30:46.120
 Like I think that is another like, you know,

30:46.120 --> 30:48.280
 sort of very interesting level of complexity.

30:48.280 --> 30:51.480
 But like if you think about just the written word, right?

30:51.480 --> 30:54.400
 Like we have, you know, we invented Papyrus

30:54.400 --> 30:56.760
 what 3000 years ago where we, you know,

30:56.760 --> 31:01.160
 you could sort of put word on paper.

31:01.160 --> 31:06.160
 And then 500 years ago, like we get the printing press

31:07.240 --> 31:11.480
 like where the word gets a little bit more ubiquitous.

31:11.480 --> 31:14.600
 And then like you really, really didn't get ubiquitous

31:14.600 --> 31:18.400
 printed word until the end of the 19th century

31:18.400 --> 31:20.720
 when the offset press was invented.

31:20.720 --> 31:22.360
 And then, you know, just sort of explodes

31:22.360 --> 31:25.360
 and like, you know, the cross product of that

31:25.360 --> 31:28.960
 and the industrial revolutions need

31:28.960 --> 31:32.880
 for educated citizens resulted in like

31:32.880 --> 31:34.720
 this rapid expansion of literacy

31:34.720 --> 31:36.000
 and the rapid expansion of the word.

31:36.000 --> 31:39.680
 But like we had 3000 years up to that point

31:39.680 --> 31:44.040
 to figure out like how to, you know, like what's,

31:44.040 --> 31:46.880
 what's journalism, what's editorial integrity?

31:46.880 --> 31:50.120
 Like what's, you know, what's scientific peer review?

31:50.120 --> 31:52.840
 And so like you built all of this mechanism

31:52.840 --> 31:57.080
 to like try to filter through all of the noise

31:57.080 --> 32:00.600
 that the technology made possible to like, you know,

32:00.600 --> 32:04.000
 sort of getting to something that society could cope with.

32:04.000 --> 32:06.600
 And like, if you think about just the piece,

32:06.600 --> 32:09.800
 the PC didn't exist 50 years ago.

32:09.800 --> 32:11.800
 And so in like this span of, you know,

32:11.800 --> 32:16.160
 like half a century, like we've gone from no digital,

32:16.160 --> 32:18.320
 you know, no ubiquitous digital technology

32:18.320 --> 32:21.080
 to like having a device that sits in your pocket

32:21.080 --> 32:23.760
 where you can sort of say whatever is on your mind

32:23.760 --> 32:26.800
 to like what would Mary have

32:26.800 --> 32:31.800
 and Mary Meeker just released her new like slide deck last week.

32:32.440 --> 32:37.360
 You know, we've got 50% penetration of the internet

32:37.360 --> 32:38.520
 to the global population.

32:38.520 --> 32:40.280
 Like there are like three and a half billion people

32:40.280 --> 32:41.720
 who are connected now.

32:41.720 --> 32:43.720
 So it's like, it's crazy, crazy.

32:43.720 --> 32:45.000
 They're like inconceivable,

32:45.000 --> 32:46.480
 like how fast all of this happened.

32:46.480 --> 32:48.720
 So, you know, it's not surprising

32:48.720 --> 32:51.000
 that we haven't figured out what to do yet,

32:51.000 --> 32:55.640
 but like we gotta really like lean into this set of problems

32:55.640 --> 33:00.200
 because like we basically have three millennia worth of work

33:00.200 --> 33:02.520
 to do about how to deal with all of this

33:02.520 --> 33:05.800
 and like probably what amounts to the next decade

33:05.800 --> 33:07.040
 worth of time.

33:07.040 --> 33:09.960
 So since we're on the topic of tough, you know,

33:09.960 --> 33:11.600
 tough challenging problems,

33:11.600 --> 33:15.200
 let's look at more on the tooling side in AI

33:15.200 --> 33:18.440
 that Microsoft is looking at as face recognition software.

33:18.440 --> 33:21.840
 So there's a lot of powerful positive use cases

33:21.840 --> 33:24.240
 for face recognition, but there's some negative ones

33:24.240 --> 33:27.200
 and we're seeing those in different governments

33:27.200 --> 33:28.160
 in the world.

33:28.160 --> 33:30.240
 So how do you, how does Microsoft think

33:30.240 --> 33:33.880
 about the use of face recognition software

33:33.880 --> 33:38.880
 as a platform in governments and companies?

33:39.400 --> 33:42.280
 Yeah, how do we strike an ethical balance here?

33:42.280 --> 33:47.280
 Yeah, I think we've articulated a clear point of view.

33:47.280 --> 33:51.840
 So Brad Smith wrote a blog post last fall,

33:51.840 --> 33:54.120
 I believe that sort of like outline,

33:54.120 --> 33:57.000
 like very specifically what, you know,

33:57.000 --> 33:59.280
 what our point of view is there.

33:59.280 --> 34:02.240
 And, you know, I think we believe that there are certain uses

34:02.240 --> 34:04.680
 to which face recognition should not be put

34:04.680 --> 34:09.160
 and we believe again that there's a need for regulation there.

34:09.160 --> 34:12.440
 Like the government should like really come in and say

34:12.440 --> 34:15.720
 that, you know, this is where the lines are.

34:15.720 --> 34:18.600
 And like we very much wanted to like figuring out

34:18.600 --> 34:20.680
 where the lines are should be a democratic process.

34:20.680 --> 34:23.240
 But in the short term, like we've drawn some lines

34:23.240 --> 34:26.640
 where, you know, we push back against uses

34:26.640 --> 34:29.440
 of face recognition technology.

34:29.440 --> 34:32.480
 You know, like this city of San Francisco, for instance,

34:32.480 --> 34:36.480
 I think has completely outlawed any government agency

34:36.480 --> 34:39.560
 from using face recognition tech.

34:39.560 --> 34:44.560
 And like that may prove to be a little bit overly broad.

34:44.560 --> 34:48.840
 But for like certain law enforcement things,

34:48.840 --> 34:53.840
 like you really, I would personally rather be overly

34:54.040 --> 34:57.400
 sort of cautious in terms of restricting use of it

34:57.400 --> 34:58.920
 until like we have, you know,

34:58.920 --> 35:02.160
 sort of defined a reasonable, you know,

35:02.160 --> 35:04.880
 democratically determined regulatory framework

35:04.880 --> 35:08.840
 for like where we could and should use it.

35:08.840 --> 35:10.880
 And, you know, the other thing there is

35:11.960 --> 35:14.000
 like we've got a bunch of research that we're doing

35:14.000 --> 35:18.400
 and a bunch of progress that we've made on bias there.

35:18.400 --> 35:20.880
 And like there are all sorts of like weird biases

35:20.880 --> 35:23.640
 that these models can have like all the way

35:23.640 --> 35:26.920
 from like the most noteworthy one where, you know,

35:26.920 --> 35:31.680
 you may have underrepresented minorities

35:31.680 --> 35:34.680
 who are like underrepresented in the training data.

35:34.680 --> 35:39.240
 And then you start learning like strange things.

35:39.240 --> 35:42.160
 But like they're even, you know, other weird things

35:42.160 --> 35:46.480
 like we've, I think we've seen in the public research

35:46.480 --> 35:49.520
 like models can learn strange things

35:49.520 --> 35:54.520
 like all doctors or men for instance.

35:54.520 --> 35:59.520
 Yeah, I mean, and so like it really is a thing where

36:00.760 --> 36:03.600
 it's very important for everybody

36:03.600 --> 36:08.440
 who is working on these things before they push publish,

36:08.440 --> 36:12.800
 they launch the experiment, they, you know, push the code

36:12.800 --> 36:17.120
 to, you know, online or they even publish the paper

36:17.120 --> 36:20.040
 that they are at least starting to think

36:20.040 --> 36:25.040
 about what some of the potential negative consequences

36:25.040 --> 36:25.880
 are some of this stuff.

36:25.880 --> 36:29.040
 I mean, this is where, you know, like the deep fake stuff

36:29.040 --> 36:32.360
 I find very worrisome just because

36:32.360 --> 36:37.360
 they're going to be some very good beneficial uses

36:39.800 --> 36:44.800
 of like GAN generated imagery.

36:46.080 --> 36:48.440
 And like, and funny enough, like one of the places

36:48.440 --> 36:52.920
 where it's actually useful is we're using the technology

36:52.920 --> 36:57.920
 right now to generate synthetic, synthetic visual data

36:58.640 --> 37:01.160
 for training some of the face recognition models

37:01.160 --> 37:03.440
 to get rid of the bias.

37:03.440 --> 37:05.800
 So like that's one like super good use of the tech,

37:05.800 --> 37:09.640
 but like, you know, it's getting good enough now

37:09.640 --> 37:12.320
 where, you know, it's going to sort of challenge

37:12.320 --> 37:15.400
 a normal human beings ability to like now you're just sort

37:15.400 --> 37:19.320
 of say like it's very expensive for someone

37:19.320 --> 37:23.280
 to fabricate a photorealistic fake video.

37:24.200 --> 37:26.920
 And like GANs are going to make it fantastically cheap

37:26.920 --> 37:30.440
 to fabricate a photorealistic fake video.

37:30.440 --> 37:33.920
 And so like what you assume you can sort of trust

37:33.920 --> 37:38.400
 is true versus like be skeptical about is about to change.

37:38.400 --> 37:40.560
 And like we're not ready for it, I don't think.

37:40.560 --> 37:42.000
 The nature of truth, right?

37:42.000 --> 37:46.360
 That's, it's also exciting because I think both you

37:46.360 --> 37:49.600
 and I probably would agree that the way to solve,

37:49.600 --> 37:52.080
 to take on that challenge is with technology.

37:52.080 --> 37:52.920
 Yeah. Right.

37:52.920 --> 37:56.800
 There's probably going to be ideas of ways to verify

37:56.800 --> 38:00.800
 which kind of video is legitimate, which kind is not.

38:00.800 --> 38:03.880
 So to me, that's an exciting possibility.

38:03.880 --> 38:07.160
 Most likely for just the comedic genius

38:07.160 --> 38:10.960
 that the internet usually creates with these kinds of videos.

38:10.960 --> 38:13.960
 And hopefully will not result in any serious harm.

38:13.960 --> 38:17.680
 Yeah. And it could be, you know, like I think

38:17.680 --> 38:22.680
 we will have technology to that may be able to detect

38:23.040 --> 38:24.440
 whether or not something's fake or real.

38:24.440 --> 38:29.440
 Although the fakes are pretty convincing

38:30.160 --> 38:34.360
 even like when you subject them to machine scrutiny.

38:34.360 --> 38:37.800
 But, you know, we also have these increasingly

38:37.800 --> 38:40.520
 interesting social networks, you know,

38:40.520 --> 38:45.520
 that are under fire right now for some of the bad things

38:45.800 --> 38:46.640
 that they do.

38:46.640 --> 38:47.720
 Like one of the things you could choose to do

38:47.720 --> 38:51.760
 with a social network is like you could,

38:51.760 --> 38:55.560
 you could use crypto and the networks

38:55.560 --> 38:59.960
 to like have content signed where you could have a like

38:59.960 --> 39:02.160
 full chain of custody that accompanied

39:02.160 --> 39:03.920
 every piece of content.

39:03.920 --> 39:06.800
 So like when you're viewing something

39:06.800 --> 39:09.640
 and like you want to ask yourself like how, you know,

39:09.640 --> 39:11.040
 how much can I trust this?

39:11.040 --> 39:12.400
 Like you can click something

39:12.400 --> 39:15.640
 and like have a verified chain of custody that shows like,

39:15.640 --> 39:19.040
 oh, this is coming from, you know, from this source.

39:19.040 --> 39:24.040
 And it's like signed by like someone whose identity I trust.

39:24.080 --> 39:25.400
 Yeah, I think having that, you know,

39:25.400 --> 39:28.040
 having that chain of custody like being able to like say,

39:28.040 --> 39:31.200
 oh, here's this video, like it may or may not

39:31.200 --> 39:33.760
 been produced using some of this deep fake technology.

39:33.760 --> 39:35.640
 But if you've got a verified chain of custody

39:35.640 --> 39:37.800
 where you can sort of trace it all the way back

39:37.800 --> 39:39.960
 to an identity and you can decide whether or not

39:39.960 --> 39:41.520
 like I trust this identity.

39:41.520 --> 39:43.360
 Like, oh no, this is really from the White House

39:43.360 --> 39:45.480
 or like this is really from the, you know,

39:45.480 --> 39:48.840
 the office of this particular presidential candidate

39:48.840 --> 39:50.960
 or it's really from, you know,

39:50.960 --> 39:55.520
 Jeff Wiener CEO of LinkedIn or Satya Nadella CEO of Microsoft.

39:55.520 --> 39:58.400
 Like that might be like one way

39:58.400 --> 39:59.960
 that you can solve some of the problems.

39:59.960 --> 40:01.800
 So like that's not the super high tech.

40:01.800 --> 40:04.480
 Like we've had all of this technology forever.

40:04.480 --> 40:06.720
 And but I think you're right.

40:06.720 --> 40:11.120
 Like it has to be some sort of technological thing

40:11.120 --> 40:15.840
 because the underlying tech that is used to create this

40:15.840 --> 40:18.800
 is not going to do anything but get better over time

40:18.800 --> 40:21.160
 and the genie is sort of out of the bottle.

40:21.160 --> 40:22.800
 There's no stuffing it back in.

40:22.800 --> 40:24.520
 And there's a social component

40:24.520 --> 40:26.600
 which I think is really healthy for democracy

40:26.600 --> 40:30.200
 where people will be skeptical about the thing they watch.

40:30.200 --> 40:31.040
 Yeah.

40:31.040 --> 40:34.160
 In general, so, you know, which is good.

40:34.160 --> 40:37.280
 Skepticism in general is good for your personal content.

40:37.280 --> 40:40.400
 So deep fakes in that sense are creating

40:40.400 --> 40:44.800
 global skepticism about can they trust what they read?

40:44.800 --> 40:46.880
 It encourages further research.

40:46.880 --> 40:48.840
 I come from the Soviet Union

40:49.800 --> 40:53.320
 where basically nobody trusted the media

40:53.320 --> 40:55.120
 because you knew it was propaganda.

40:55.120 --> 40:59.160
 And that kind of skepticism encouraged further research

40:59.160 --> 41:02.360
 about ideas supposed to just trusting anyone's source.

41:02.360 --> 41:05.440
 Well, like I think it's one of the reasons why the,

41:05.440 --> 41:09.440
 you know, the scientific method and our apparatus

41:09.440 --> 41:11.480
 of modern science is so good.

41:11.480 --> 41:15.360
 Like because you don't have to trust anything.

41:15.360 --> 41:18.520
 Like you, like the whole notion of, you know,

41:18.520 --> 41:21.320
 like modern science beyond the fact that, you know,

41:21.320 --> 41:23.440
 this is a hypothesis and this is an experiment

41:23.440 --> 41:24.840
 to test the hypothesis.

41:24.840 --> 41:27.360
 And, you know, like this is a peer review process

41:27.360 --> 41:30.080
 for scrutinizing published results.

41:30.080 --> 41:33.280
 But like stuff's also supposed to be reproducible.

41:33.280 --> 41:35.240
 So like, you know, it's been vetted by this process,

41:35.240 --> 41:38.000
 but like you also are expected to publish enough detail

41:38.000 --> 41:41.480
 where, you know, if you are sufficiently skeptical

41:41.480 --> 41:44.720
 of the thing, you can go try to like reproduce it yourself.

41:44.720 --> 41:47.560
 And like, I don't know what it is.

41:47.560 --> 41:49.920
 Like, I think a lot of engineers are like this

41:49.920 --> 41:52.600
 where like, you know, sort of this, like your brain

41:52.600 --> 41:55.520
 is sort of wired for skepticism.

41:55.520 --> 41:58.000
 Like you don't just first order trust everything

41:58.000 --> 42:00.040
 that you see and encounter.

42:00.040 --> 42:02.560
 And like you're sort of curious to understand,

42:02.560 --> 42:04.480
 you know, the next thing.

42:04.480 --> 42:09.080
 But like, I think it's an entirely healthy thing.

42:09.080 --> 42:12.280
 And like we need a little bit more of that right now.

42:12.280 --> 42:16.200
 So I'm not a large business owner.

42:16.200 --> 42:23.200
 So I'm just, I'm just a huge fan of many of Microsoft products.

42:23.200 --> 42:25.360
 I mean, I still, actually in terms of,

42:25.360 --> 42:27.000
 I generate a lot of graphics and images

42:27.000 --> 42:28.640
 and I still use PowerPoint to do that.

42:28.640 --> 42:30.440
 It beats Illustrator for me.

42:30.440 --> 42:34.480
 Even professional sort of, it's fascinating.

42:34.480 --> 42:39.560
 So I wonder what is the future of, let's say,

42:39.560 --> 42:41.920
 windows and office look like?

42:41.920 --> 42:43.840
 Is do you see it?

42:43.840 --> 42:45.880
 I mean, I remember looking forward to XP.

42:45.880 --> 42:48.200
 Was it exciting when XP was released?

42:48.200 --> 42:51.080
 Just like you said, I don't remember when 95 was released.

42:51.080 --> 42:53.800
 But XP for me was a big celebration.

42:53.800 --> 42:56.000
 And when 10 came out, I was like,

42:56.000 --> 42:58.040
 okay, well, it's nice, it's a nice improvement.

42:58.040 --> 43:02.600
 But so what do you see the future of these products?

43:02.600 --> 43:04.640
 You know, I think there's a bunch of excitement.

43:04.640 --> 43:07.160
 I mean, on the office front,

43:07.160 --> 43:13.440
 there's going to be this like increasing productivity

43:13.440 --> 43:17.080
 wins that are coming out of some of these AI powered features

43:17.080 --> 43:19.000
 that are coming, like the products will sort of get

43:19.000 --> 43:21.120
 smarter and smarter in like a very subtle way.

43:21.120 --> 43:24.120
 Like there's not going to be this big bang moment

43:24.120 --> 43:27.080
 where, you know, like Clippy is going to reemerge

43:27.080 --> 43:27.960
 and it's going to be...

43:27.960 --> 43:28.680
 Wait a minute.

43:28.680 --> 43:30.520
 Okay, well, I have to wait, wait, wait.

43:30.520 --> 43:31.960
 It's Clippy coming back.

43:31.960 --> 43:34.560
 Well, quite seriously.

43:34.560 --> 43:37.920
 So injection of AI, there's not much,

43:37.920 --> 43:39.040
 or at least I'm not familiar,

43:39.040 --> 43:41.200
 sort of assistive type of stuff going on

43:41.200 --> 43:43.600
 inside the office products,

43:43.600 --> 43:47.600
 like a Clippy style assistant, personal assistant.

43:47.600 --> 43:50.560
 Do you think that there's a possibility

43:50.560 --> 43:52.000
 of that in the future?

43:52.000 --> 43:54.680
 So I think there are a bunch of like very small ways

43:54.680 --> 43:57.320
 in which like machine learning power

43:57.320 --> 44:00.080
 and assistive things are in the product right now.

44:00.080 --> 44:04.800
 So there are a bunch of interesting things,

44:04.800 --> 44:09.280
 like the auto response stuff's getting better and better

44:09.280 --> 44:12.160
 and it's like getting to the point where, you know,

44:12.160 --> 44:14.960
 it can auto respond with like, okay,

44:14.960 --> 44:19.080
 let this person is clearly trying to schedule a meeting

44:19.080 --> 44:21.520
 so it looks at your calendar and it automatically

44:21.520 --> 44:24.080
 like tries to find like a time and a space

44:24.080 --> 44:26.240
 that's mutually interesting.

44:26.240 --> 44:31.240
 Like we have this notion of Microsoft search

44:33.520 --> 44:34.960
 where it's like not just web search,

44:34.960 --> 44:38.200
 but it's like search across like all of your information

44:38.200 --> 44:43.200
 that's sitting inside of like your Office 365 tenant

44:43.320 --> 44:46.880
 and like, you know, potentially in other products.

44:46.880 --> 44:49.680
 And like we have this thing called the Microsoft Graph

44:49.680 --> 44:53.400
 that is basically a API federator that, you know,

44:53.400 --> 44:57.960
 sort of like gets you hooked up across the entire breadth

44:57.960 --> 44:59.760
 of like all of the, you know,

44:59.760 --> 45:01.640
 like what were information silos

45:01.640 --> 45:04.720
 before they got woven together with the graph.

45:05.680 --> 45:07.880
 Like that is like getting increasing

45:07.880 --> 45:09.160
 with increasing effectiveness,

45:09.160 --> 45:11.280
 sort of plumbed into the,

45:11.280 --> 45:13.120
 into some of these auto response things

45:13.120 --> 45:15.840
 where you're going to be able to see the system

45:15.840 --> 45:18.200
 like automatically retrieve information for you.

45:18.200 --> 45:21.160
 Like if, you know, like I frequently send out,

45:21.160 --> 45:24.080
 you know, emails to folks where like I can't find a paper

45:24.080 --> 45:25.400
 or a document or whatnot.

45:25.400 --> 45:26.840
 There's no reason why the system won't be able

45:26.840 --> 45:27.680
 to do that for you.

45:27.680 --> 45:29.560
 And like, I think the,

45:29.560 --> 45:33.640
 it's building towards like having things that look more

45:33.640 --> 45:37.880
 like like a fully integrated, you know, assistant,

45:37.880 --> 45:40.720
 but like you'll have a bunch of steps

45:40.720 --> 45:42.800
 that you will see before you,

45:42.800 --> 45:45.120
 like it will not be this like big bang thing

45:45.120 --> 45:47.400
 where like Clippy comes back and you've got this like,

45:47.400 --> 45:49.360
 you know, manifestation of, you know,

45:49.360 --> 45:52.000
 like a fully, fully powered assistant.

45:53.320 --> 45:56.920
 So I think that's, that's definitely coming out.

45:56.920 --> 45:58.680
 Like all of the, you know, collaboration,

45:58.680 --> 46:00.720
 co authoring stuff's getting better.

46:00.720 --> 46:02.200
 You know, it's like really interesting.

46:02.200 --> 46:07.200
 Like if you look at how we use the office product portfolio

46:08.320 --> 46:10.840
 at Microsoft, like more and more of it is happening

46:10.840 --> 46:14.480
 inside of like teams as a canvas.

46:14.480 --> 46:17.160
 And like it's this thing where, you know,

46:17.160 --> 46:19.840
 that you've got collaboration is like

46:19.840 --> 46:21.560
 at the center of the product.

46:21.560 --> 46:26.560
 And like we, we, we built some like really cool stuff

46:26.720 --> 46:29.440
 that's some of, which is about to be open source

46:29.440 --> 46:33.120
 that are sort of framework level things for doing,

46:33.120 --> 46:35.600
 for doing co authoring.

46:35.600 --> 46:36.440
 That's awesome.

46:36.440 --> 46:38.920
 So in, is there a cloud component to that?

46:38.920 --> 46:41.880
 So on the web or is it,

46:41.880 --> 46:43.640
 forgive me if I don't already know this,

46:43.640 --> 46:45.600
 but with office 365,

46:45.600 --> 46:48.480
 we still, the collaboration we do, if we're doing Word,

46:48.480 --> 46:50.640
 we're still sending the file around.

46:50.640 --> 46:51.480
 No, no, no, no.

46:51.480 --> 46:53.400
 So this is,

46:53.400 --> 46:55.240
 we're already a little bit better than that.

46:55.240 --> 46:57.360
 And like, you know, so like the fact that you're unaware

46:57.360 --> 46:59.120
 of it means we've got a better job to do,

46:59.120 --> 47:01.960
 like helping you discover, discover this stuff.

47:02.880 --> 47:06.360
 But yeah, I mean, it's already like got a huge,

47:06.360 --> 47:07.200
 huge cloud component.

47:07.200 --> 47:09.680
 And like part of, you know, part of this framework stuff,

47:09.680 --> 47:12.640
 I think we're calling it, like I,

47:12.640 --> 47:14.520
 like we've been working on it for a couple of years.

47:14.520 --> 47:17.200
 So like, I know the, the internal OLA code name for it,

47:17.200 --> 47:18.640
 but I think when we launched it to build,

47:18.640 --> 47:20.720
 it's called the fluid framework.

47:21.920 --> 47:25.080
 And, but like what fluid lets you do is like,

47:25.080 --> 47:27.920
 you can go into a conversation that you're having in teams

47:27.920 --> 47:30.280
 and like reference, like part of a spreadsheet

47:30.280 --> 47:32.600
 that you're working on,

47:32.600 --> 47:35.600
 where somebody's like sitting in the Excel canvas,

47:35.600 --> 47:37.760
 like working on the spreadsheet with a, you know,

47:37.760 --> 47:39.120
 charter whatnot.

47:39.120 --> 47:42.000
 And like, you can sort of embed like part of the spreadsheet

47:42.000 --> 47:43.240
 in the team's conversation,

47:43.240 --> 47:46.520
 where like you can dynamically update in like all

47:46.520 --> 47:49.400
 of the changes that you're making to the,

47:49.400 --> 47:51.280
 to this object or like, you know,

47:51.280 --> 47:54.680
 coordinate and everything is sort of updating in real time.

47:54.680 --> 47:58.000
 So like you can be in whatever canvas is most convenient

47:58.000 --> 48:00.400
 for you to get your work done.

48:00.400 --> 48:03.400
 So out of my own sort of curiosity as an engineer,

48:03.400 --> 48:06.280
 I know what it's like to sort of lead a team

48:06.280 --> 48:08.280
 of 10, 15 engineers.

48:08.280 --> 48:11.680
 Microsoft has, I don't know what the numbers are,

48:11.680 --> 48:14.920
 maybe 15, maybe 60,000 engineers, maybe 40.

48:14.920 --> 48:16.160
 I don't know exactly what the number is.

48:16.160 --> 48:17.000
 It's a lot.

48:17.000 --> 48:18.520
 It's tens of thousands.

48:18.520 --> 48:20.640
 Right. This is more than 10 or 15.

48:23.640 --> 48:28.640
 I mean, you've led different sizes,

48:28.720 --> 48:30.560
 mostly large sizes of engineers.

48:30.560 --> 48:33.840
 What does it take to lead such a large group

48:33.840 --> 48:37.480
 into a continue innovation,

48:37.480 --> 48:40.240
 continue being highly productive

48:40.240 --> 48:43.200
 and yet develop all kinds of new ideas

48:43.200 --> 48:45.120
 and yet maintain like, what does it take

48:45.120 --> 48:49.000
 to lead such a large group of brilliant people?

48:49.000 --> 48:52.080
 I think the thing that you learn

48:52.080 --> 48:55.120
 as you manage larger and larger scale

48:55.120 --> 48:57.920
 is that there are three things

48:57.920 --> 49:00.480
 that are like very, very important

49:00.480 --> 49:02.360
 for big engineering teams.

49:02.360 --> 49:06.320
 Like one is like having some sort of forethought

49:06.320 --> 49:09.840
 about what it is that you're going to be building

49:09.840 --> 49:11.040
 over large periods of time.

49:11.040 --> 49:11.880
 Like not exactly.

49:11.880 --> 49:13.760
 Like you don't need to know that like,

49:13.760 --> 49:16.440
 I'm putting all my chips on this one product

49:16.440 --> 49:17.760
 and like this is going to be the thing.

49:17.760 --> 49:21.440
 But it's useful to know what sort of capabilities

49:21.440 --> 49:23.080
 you think you're going to need to have

49:23.080 --> 49:24.720
 to build the products of the future

49:24.720 --> 49:28.000
 and then like invest in that infrastructure.

49:28.000 --> 49:31.520
 Like whether, and I'm not just talking about storage systems

49:31.520 --> 49:33.480
 or cloud APIs, it's also like,

49:33.480 --> 49:35.360
 what is your development process look like?

49:35.360 --> 49:36.720
 What tools do you want?

49:36.720 --> 49:39.560
 Like what culture do you want to build

49:39.560 --> 49:42.760
 around like how you're sort of collaborating together

49:42.760 --> 49:45.720
 to like make complicated technical things?

49:45.720 --> 49:48.080
 And so like having an opinion and investing in that

49:48.080 --> 49:50.480
 is like, it just gets more and more important.

49:50.480 --> 49:54.520
 And like the sooner you can get a concrete set of opinions,

49:54.520 --> 49:57.680
 like the better you're going to be.

49:57.680 --> 50:01.600
 Like you can wing it for a while at small scales.

50:01.600 --> 50:03.160
 Like, you know, when you start a company,

50:03.160 --> 50:06.320
 like you don't have to be like super specific about it.

50:06.320 --> 50:10.000
 But like the biggest miseries that I've ever seen

50:10.000 --> 50:12.640
 as an engineering leader are in places

50:12.640 --> 50:14.440
 where you didn't have a clear enough opinion

50:14.440 --> 50:16.800
 about those things soon enough.

50:16.800 --> 50:20.240
 And then you just sort of go create a bunch of technical debt

50:20.240 --> 50:24.000
 and like culture debt that is excruciatingly painful

50:24.000 --> 50:25.760
 to clean up.

50:25.760 --> 50:28.640
 So like that's one bundle of things.

50:28.640 --> 50:33.640
 Like the other, you know, another bundle of things is

50:33.640 --> 50:37.440
 like it's just really, really important to

50:38.960 --> 50:43.960
 like have a clear mission that's not just some cute crap

50:45.520 --> 50:48.880
 you say because like you think you should have a mission,

50:48.880 --> 50:52.880
 but like something that clarifies for people

50:52.880 --> 50:55.680
 like where it is that you're headed together.

50:57.160 --> 50:58.520
 Like I know it's like probably

50:58.520 --> 51:00.320
 like a little bit too popular right now,

51:00.320 --> 51:05.320
 but Yval Harari's book, Sapiens,

51:07.240 --> 51:12.240
 one of the central ideas in his book is that

51:12.440 --> 51:16.840
 like storytelling is like the quintessential thing

51:16.840 --> 51:20.480
 for coordinating the activities of large groups of people.

51:20.480 --> 51:22.320
 Like once you get past Dunbar's number

51:23.360 --> 51:25.800
 and like I've really, really seen that

51:25.800 --> 51:27.320
 just managing engineering teams.

51:27.320 --> 51:32.080
 Like you can just brute force things

51:32.080 --> 51:35.160
 when you're less than 120, 150 folks

51:35.160 --> 51:37.520
 where you can sort of know and trust

51:37.520 --> 51:40.920
 and understand what the dynamics are between all the people.

51:40.920 --> 51:41.840
 But like past that,

51:41.840 --> 51:45.440
 like things just sort of start to catastrophically fail

51:45.440 --> 51:48.760
 if you don't have some sort of set of shared goals

51:48.760 --> 51:50.480
 that you're marching towards.

51:50.480 --> 51:52.960
 And so like even though it sounds touchy feely

51:52.960 --> 51:55.640
 and you know, like a bunch of technical people

51:55.640 --> 51:58.200
 will sort of balk at the idea that like you need

51:58.200 --> 52:01.680
 to like have a clear, like the missions

52:01.680 --> 52:03.560
 like very, very, very important.

52:03.560 --> 52:04.640
 Yval's right, right?

52:04.640 --> 52:07.520
 Stories, that's how our society,

52:07.520 --> 52:09.360
 that's the fabric that connects us all of us

52:09.360 --> 52:11.120
 is these powerful stories.

52:11.120 --> 52:13.440
 And that works for companies too, right?

52:13.440 --> 52:14.520
 It works for everything.

52:14.520 --> 52:16.520
 Like I mean, even down to like, you know,

52:16.520 --> 52:18.280
 you sort of really think about like our currency

52:18.280 --> 52:19.960
 for instance is a story.

52:19.960 --> 52:23.360
 Our constitution is a story, our laws are story.

52:23.360 --> 52:27.840
 I mean, like we believe very, very, very strongly in them

52:27.840 --> 52:29.960
 and thank God we do.

52:29.960 --> 52:33.040
 But like they are, they're just abstract things.

52:33.040 --> 52:34.000
 Like they're just words.

52:34.000 --> 52:36.520
 Like if we don't believe in them, they're nothing.

52:36.520 --> 52:39.440
 And in some sense, those stories are platforms

52:39.440 --> 52:43.040
 and the kinds some of which Microsoft is creating, right?

52:43.040 --> 52:46.360
 Yeah, platforms in which we define the future.

52:46.360 --> 52:48.600
 So last question, what do you,

52:48.600 --> 52:50.080
 let's get philosophical maybe,

52:50.080 --> 52:51.480
 bigger than even Microsoft.

52:51.480 --> 52:56.280
 What do you think the next 2030 plus years

52:56.280 --> 53:00.120
 looks like for computing, for technology, for devices?

53:00.120 --> 53:03.760
 Do you have crazy ideas about the future of the world?

53:04.600 --> 53:06.400
 Yeah, look, I think we, you know,

53:06.400 --> 53:09.480
 we're entering this time where we've got,

53:10.640 --> 53:13.360
 we have technology that is progressing

53:13.360 --> 53:15.800
 at the fastest rate that it ever has.

53:15.800 --> 53:20.800
 And you've got, you get some really big social problems

53:20.800 --> 53:25.800
 like society scale problems that we have to tackle.

53:26.320 --> 53:28.720
 And so, you know, I think we're gonna rise to the challenge

53:28.720 --> 53:30.560
 and like figure out how to intersect

53:30.560 --> 53:32.400
 like all of the power of this technology

53:32.400 --> 53:35.320
 with all of the big challenges that are facing us,

53:35.320 --> 53:37.840
 whether it's, you know, global warming,

53:37.840 --> 53:41.000
 whether it's like the biggest remainder of the population

53:41.000 --> 53:46.000
 boom is in Africa for the next 50 years or so.

53:46.800 --> 53:49.360
 And like global warming is gonna make it increasingly

53:49.360 --> 53:52.600
 difficult to feed the global population in particular,

53:52.600 --> 53:54.200
 like in this place where you're gonna have

53:54.200 --> 53:56.600
 like the biggest population boom.

53:57.720 --> 54:01.520
 I think we, you know, like AI is gonna,

54:01.520 --> 54:03.560
 like if we push it in the right direction,

54:03.560 --> 54:05.680
 like it can do like incredible things

54:05.680 --> 54:10.160
 to empower all of us to achieve our full potential

54:10.160 --> 54:15.160
 and to, you know, like live better lives.

54:15.160 --> 54:20.160
 But like that also means focus on like

54:20.520 --> 54:22.040
 some super important things,

54:22.040 --> 54:23.960
 like how can you apply it to healthcare

54:23.960 --> 54:28.960
 to make sure that, you know, like our quality and cost of,

54:29.640 --> 54:32.080
 and sort of ubiquity of health coverage

54:32.080 --> 54:35.080
 is better and better over time.

54:35.080 --> 54:37.960
 Like that's more and more important every day

54:37.960 --> 54:40.880
 is like in the United States

54:40.880 --> 54:43.280
 and like the rest of the industrialized world.

54:43.280 --> 54:45.720
 So Western Europe, China, Japan, Korea,

54:45.720 --> 54:48.880
 like you've got this population bubble

54:48.880 --> 54:52.880
 of like aging working, you know, working age folks

54:52.880 --> 54:56.200
 who are, you know, at some point over the next 20, 30 years

54:56.200 --> 54:58.000
 they're gonna be largely retired

54:58.000 --> 55:00.160
 and like you're gonna have more retired people

55:00.160 --> 55:01.200
 than working age people.

55:01.200 --> 55:02.520
 And then like you've got, you know,

55:02.520 --> 55:04.800
 sort of natural questions about who's gonna take care

55:04.800 --> 55:07.120
 of all the old folks and who's gonna do all the work.

55:07.120 --> 55:11.040
 And the answers to like all of these sorts of questions

55:11.040 --> 55:13.200
 like where you're sort of running into, you know,

55:13.200 --> 55:16.080
 like constraints of the, you know,

55:16.080 --> 55:20.080
 the world and of society has always been like

55:20.080 --> 55:23.000
 what tech is gonna like help us get around this.

55:23.000 --> 55:26.360
 You know, like when I was a kid in the 70s and 80s,

55:26.360 --> 55:29.800
 like we talked all the time about like population boom,

55:29.800 --> 55:32.200
 population boom, like we're gonna,

55:32.200 --> 55:34.360
 like we're not gonna be able to like feed the planet.

55:34.360 --> 55:36.800
 And like we were like right in the middle

55:36.800 --> 55:38.200
 of the green revolution

55:38.200 --> 55:43.200
 where like this massive technology driven increase

55:44.560 --> 55:47.520
 and crop productivity like worldwide.

55:47.520 --> 55:49.320
 And like some of that was like taking some of the things

55:49.320 --> 55:52.560
 that we knew in the West and like getting them distributed

55:52.560 --> 55:55.760
 to the, you know, to the developing world.

55:55.760 --> 55:59.360
 And like part of it were things like, you know,

55:59.360 --> 56:03.280
 just smarter biology like helping us increase.

56:03.280 --> 56:06.760
 And like we don't talk about like, yeah,

56:06.760 --> 56:10.320
 overpopulation anymore because like we can more or less,

56:10.320 --> 56:12.000
 we sort of figured out how to feed the world.

56:12.000 --> 56:14.760
 Like that's a technology story.

56:14.760 --> 56:19.480
 And so like I'm super, super hopeful about the future

56:19.480 --> 56:24.080
 and in the ways where we will be able to apply technology

56:24.080 --> 56:28.040
 to solve some of these super challenging problems.

56:28.040 --> 56:31.360
 Like I've, like one of the things

56:31.360 --> 56:34.680
 that I'm trying to spend my time doing right now

56:34.680 --> 56:36.600
 is trying to get everybody else to be hopeful

56:36.600 --> 56:38.720
 as well because, you know, back to Harari,

56:38.720 --> 56:41.160
 like we are the stories that we tell.

56:41.160 --> 56:44.320
 Like if we, you know, if we get overly pessimistic right now

56:44.320 --> 56:49.320
 about like the potential future of technology, like we,

56:49.320 --> 56:53.680
 you know, like we may fail to fail to get all the things

56:53.680 --> 56:56.880
 in place that we need to like have our best possible future.

56:56.880 --> 56:59.440
 And that kind of hopeful optimism.

56:59.440 --> 57:03.160
 I'm glad that you have it because you're leading large groups

57:03.160 --> 57:05.600
 of engineers that are actually defining

57:05.600 --> 57:06.720
 that are writing that story,

57:06.720 --> 57:08.320
 that are helping build that future,

57:08.320 --> 57:10.000
 which is super exciting.

57:10.000 --> 57:12.320
 And I agree with everything you said,

57:12.320 --> 57:14.840
 except I do hope Clippy comes back.

57:16.400 --> 57:17.760
 We miss him.

57:17.760 --> 57:19.360
 I speak for the people.

57:19.360 --> 57:21.800
 So, Kellen, thank you so much for talking to me.

57:21.800 --> 57:22.640
 Thank you so much for having me.

57:22.640 --> 57:43.640
 It was a pleasure.