Kevin Scott: Microsoft CTO
技术与编程商业与创业AI 与机器学习政治与社会音乐与艺术
📋 章节目录
暂无章节信息
🔑 关键词
microsoftgoingdontechnologyplatforminterestinggotdatabetterdoingfolksgettingbunchstuffsuperworkingresearchhavingfuturecompany
💬 精彩语录
暂无语录
🎙️ 完整对话（1161 条）
Lex Fridman (00:00.000)
The following is a conversation with Kevin Scott,
以下是与凯文·斯科特的对话，
Lex Fridman (00:03.440)
the CTO of Microsoft.
微软首席技术官。
Lex Fridman (00:06.080)
Before that, he was the senior vice president
在此之前，他曾担任高级副总裁
Lex Fridman (00:08.540)
of engineering and operations at LinkedIn.
LinkedIn 的工程和运营部门。
Lex Fridman (00:11.080)
And before that, he oversaw mobile ads engineering
在此之前，他负责监督移动广告工程
Kevin Scott (00:14.160)
at Google.
在谷歌。
Lex Fridman (00:15.960)
He also has a podcast called Behind the Tech
他还有一个名为“Behind the Tech”的播客
Kevin Scott (00:18.960)
with Kevin Scott, which I'm a fan of.
和凯文·斯科特，我是他的粉丝。
Lex Fridman (00:21.860)
This was a fun and wide ranging conversation
这是一次有趣且范围广泛的对话
Kevin Scott (00:24.240)
that covered many aspects of computing.
这涵盖了计算的许多方面。
Lex Fridman (00:26.680)
It happened over a month ago,
事情发生在一个多月前
Kevin Scott (00:28.800)
before the announcement of Microsoft's investment
在微软宣布投资之前
Lex Fridman (00:30.960)
in OpenAI that a few people have asked me about.
有几个人问过我有关 OpenAI 的问题。
Kevin Scott (00:34.400)
I'm sure there'll be one or two people in the future
我确信未来还会有一两个人
Lex Fridman (00:38.080)
that'll talk with me about the impact of that investment.
那将与我讨论该投资的影响。
Kevin Scott (00:42.240)
This is the Artificial Intelligence Podcast.
这是人工智能播客。
Lex Fridman (00:45.400)
If you enjoy it, subscribe on YouTube,
如果您喜欢，请在 YouTube 上订阅，
Kevin Scott (00:47.640)
give it five stars on iTunes,
在 iTunes 上给它五颗星，
Lex Fridman (00:49.400)
support it on Patreon,
在 Patreon 上支持它，
Kevin Scott (00:50.920)
or simply connect with me on Twitter at Lex Friedman,
或者直接在 Twitter 上联系我：Lex Friedman，
Lex Fridman (00:54.240)
spelled F R I D M A N.
Lex Fridman (00:57.640)
And I'd like to give a special thank you
Lex Fridman (00:59.200)
to Tom and Nelante Bighousen
Kevin Scott (01:01.920)
for their support of the podcast on Patreon.
Lex Fridman (01:04.560)
Thanks Tom and Nelante.
Kevin Scott (01:06.040)
Hope I didn't mess up your last name too bad.
Lex Fridman (01:08.340)
Your support means a lot
Lex Fridman (01:10.480)
and inspires me to keep this series going.
Lex Fridman (01:13.440)
And now, here's my conversation with Kevin Scott.
Kevin Scott (01:18.100)
You've described yourself as a kid in a candy store
Lex Fridman (01:20.680)
at Microsoft because of all the interesting projects
Kevin Scott (01:22.920)
that are going on.
Lex Fridman (01:24.160)
Can you try to do the impossible task
Lex Fridman (01:27.920)
and give a brief whirlwind view
Lex Fridman (01:31.720)
of all the spaces that Microsoft is working in?
Lex Fridman (01:34.500)
Both research and product?
Lex Fridman (01:37.400)
If you include research,
Kevin Scott (01:38.360)
it becomes even more difficult.
Lex Fridman (01:46.840)
I think broadly speaking,
Kevin Scott (01:48.800)
Microsoft's product portfolio includes everything
Lex Fridman (01:53.680)
from big cloud business,
Kevin Scott (01:56.880)
like a big set of SaaS services.
Lex Fridman (01:59.320)
We have sort of the original,
Kevin Scott (02:01.920)
or like some of what are among the original
Lex Fridman (02:05.520)
productivity software products that everybody uses.
Kevin Scott (02:09.560)
We have an operating system business.
Lex Fridman (02:11.160)
We have a hardware business where we make everything
Kevin Scott (02:14.540)
from computer mice and headphones
Lex Fridman (02:18.400)
to high end personal computers and laptops.
Kevin Scott (02:23.400)
We have a fairly broad ranging research group
Lex Fridman (02:27.640)
where we have people doing everything
Kevin Scott (02:29.640)
from economics research.
Lex Fridman (02:31.840)
So there's this really, really smart young economist,
Kevin Scott (02:35.880)
Glenn Weil, who my group works with a lot,
Lex Fridman (02:39.720)
who's doing this research on these things
Kevin Scott (02:42.840)
called radical markets.
Lex Fridman (02:45.120)
He's written an entire technical book
Kevin Scott (02:48.080)
about this whole notion of radical markets.
Lex Fridman (02:51.080)
So like the research group sort of spans from that
Kevin Scott (02:53.480)
to human computer interaction to artificial intelligence.
Lex Fridman (02:56.800)
And we have GitHub, we have LinkedIn,
Kevin Scott (03:01.000)
we have a search advertising and news business
Lex Fridman (03:05.760)
and like probably a bunch of stuff
Kevin Scott (03:07.320)
that I'm embarrassingly not recounting in this list.
Lex Fridman (03:11.240)
Gaming to Xbox and so on, right?
Kevin Scott (03:12.840)
Yeah, gaming for sure.
Lex Fridman (03:14.080)
Like I was having a super fun conversation this morning
Kevin Scott (03:17.880)
with Phil Spencer.
Lex Fridman (03:19.480)
So when I was in college,
Kevin Scott (03:21.260)
there was this game that LucasArts made
Lex Fridman (03:25.560)
called Day of the Tentacle
Kevin Scott (03:27.600)
that my friends and I played forever.
Lex Fridman (03:30.160)
And like we're doing some interesting collaboration now
Kevin Scott (03:33.920)
with the folks who made Day of the Tentacle.
Lex Fridman (03:37.920)
And I was like completely nerding out with Tim Schafer,
Kevin Scott (03:40.840)
like the guy who wrote a Day of the Tentacle this morning,
Lex Fridman (03:43.880)
just a complete fan boy,
Kevin Scott (03:45.800)
which sort of it like happens a lot.
Lex Fridman (03:49.880)
Like Microsoft has been doing so much stuff
Kevin Scott (03:53.320)
at such breadth for such a long period of time
Lex Fridman (03:56.000)
that like being CTO like most of the time,
Kevin Scott (04:00.880)
my job is very, very serious.
Lex Fridman (04:02.200)
And sometimes like I get caught up
Kevin Scott (04:05.620)
in like how amazing it is to be able to have
Lex Fridman (04:10.620)
the conversations that I have with the people
Kevin Scott (04:12.800)
I get to have them with.
Lex Fridman (04:14.640)
Yeah, to reach back into the sentimental.
Lex Fridman (04:17.080)
And what's the radical markets and the economics?
Lex Fridman (04:21.640)
So the idea with radical markets is like,
Lex Fridman (04:24.760)
can you come up with new market based mechanisms to,
Lex Fridman (04:32.320)
you know, I think we have this,
Kevin Scott (04:33.840)
we're having this debate right now,
Lex Fridman (04:35.240)
like does capitalism work like free markets work?
Kevin Scott (04:40.040)
Can the incentive structures
Lex Fridman (04:43.000)
that are built into these systems produce outcomes
Kevin Scott (04:46.320)
that are creating sort of equitably distributed benefits
Lex Fridman (04:51.520)
for every member of society?
Kevin Scott (04:55.360)
You know, and I think it's a reasonable,
Lex Fridman (04:56.960)
reasonable set of questions to be asking.
Lex Fridman (04:59.520)
And so what Glenn, and so like, you know,
Lex Fridman (05:02.120)
one mode of thought there,
Kevin Scott (05:03.120)
like if you have doubts that the markets
Lex Fridman (05:05.920)
are actually working, you can sort of like tip towards
Kevin Scott (05:08.360)
like, okay, let's become more socialist
Lex Fridman (05:10.760)
and, you know, like have central planning and, you know,
Kevin Scott (05:13.640)
governments or some other central organization
Lex Fridman (05:15.760)
is like making a bunch of decisions
Kevin Scott (05:18.240)
about how, you know, sort of work gets done
Lex Fridman (05:22.000)
and, you know, like where the, you know,
Kevin Scott (05:24.520)
where the investments and where the outputs
Lex Fridman (05:26.360)
of those investments get distributed.
Kevin Scott (05:28.840)
Glenn's notion is like, lean more
Lex Fridman (05:32.120)
into like the market based mechanism.
Lex Fridman (05:35.760)
So like, for instance, you know,
Lex Fridman (05:37.840)
this is one of the more radical ideas,
Kevin Scott (05:39.540)
like suppose that you had a radical pricing mechanism
Lex Fridman (05:45.140)
for assets like real estate where you were,
Kevin Scott (05:50.560)
you could be bid out of your position
Lex Fridman (05:53.560)
in your home, you know, for instance.
Lex Fridman (05:58.680)
So like if somebody came along and said,
Lex Fridman (06:01.080)
you know, like I can find higher economic utility
Kevin Scott (06:04.380)
for this piece of real estate
Lex Fridman (06:05.720)
that you're running your business in,
Kevin Scott (06:08.680)
like then like you either have to, you know,
Lex Fridman (06:13.040)
sort of bid to sort of stay
Kevin Scott (06:16.480)
or like the thing that's got the higher economic utility,
Lex Fridman (06:19.960)
you know, sort of takes over the asset
Kevin Scott (06:21.480)
which would make it very difficult
Lex Fridman (06:23.700)
to have the same sort of rent seeking behaviors
Kevin Scott (06:27.580)
that you've got right now
Lex Fridman (06:29.000)
because like if you did speculative bidding,
Kevin Scott (06:34.000)
like you would very quickly like lose a whole lot of money.
Lex Fridman (06:40.440)
And so like the prices of the assets
Kevin Scott (06:42.380)
would be sort of like very closely indexed
Lex Fridman (06:45.640)
to like the value that they could produce.
Lex Fridman (06:49.720)
And like, because like you'd have this sort
Lex Fridman (06:52.120)
of real time mechanism that would force you
Kevin Scott (06:53.940)
to sort of mark the value of the asset to the market,
Lex Fridman (06:56.800)
then it could be taxed appropriately.
Kevin Scott (06:58.560)
Like you couldn't sort of sit on this thing and say,
Lex Fridman (07:00.400)
oh, like this house is only worth 10,000 bucks
Kevin Scott (07:03.040)
when like everything around it is worth 10 million.
Lex Fridman (07:06.620)
That's really, so it's an incentive structure
Kevin Scott (07:08.720)
that where the prices match the value much better.
Lex Fridman (07:13.160)
Yeah, and Glenn does a much better job than I do
Kevin Scott (07:16.360)
at selling and I probably picked the world's worst example,
Lex Fridman (07:18.960)
you know, and it's intentionally provocative,
Lex Fridman (07:24.560)
so like this whole notion,
Lex Fridman (07:25.800)
like I'm not sure whether I like this notion
Kevin Scott (07:28.920)
that like we can have a set of market mechanisms
Lex Fridman (07:31.120)
where I could get bid out of my property, you know,
Lex Fridman (07:35.360)
but you know, like if you're thinking about something
Lex Fridman (07:37.680)
like Elizabeth Warren's wealth tax, for instance,
Kevin Scott (07:42.480)
like you would have, I mean, it'd be really interesting
Lex Fridman (07:45.600)
in like how you would actually set the price on the assets
Lex Fridman (07:50.100)
and like you might have to have a mechanism like that
Lex Fridman (07:52.040)
if you put a tax like that in place.
Kevin Scott (07:54.160)
It's really interesting that that kind of research,
Lex Fridman (07:56.440)
at least tangentially is touching Microsoft research.
Kevin Scott (08:00.280)
That you're really thinking broadly.
Lex Fridman (08:02.560)
Maybe you can speak to, this connects to AI,
Lex Fridman (08:08.360)
so we have a candidate, Andrew Yang,
Lex Fridman (08:10.640)
who kind of talks about artificial intelligence
Lex Fridman (08:13.440)
and the concern that people have about, you know,
Lex Fridman (08:16.620)
automation's impact on society and arguably,
Kevin Scott (08:19.920)
Microsoft is at the cutting edge of innovation
Lex Fridman (08:23.340)
in all these kinds of ways and so it's pushing AI forward.
Lex Fridman (08:27.040)
How do you think about combining all our conversations
Lex Fridman (08:30.000)
together here with radical markets and socialism
Lex Fridman (08:32.840)
and innovation in AI that Microsoft is doing
Lex Fridman (08:37.500)
and then Andrew Yang's worry that that will result
Kevin Scott (08:44.560)
in job loss for the lower and so on.
Lex Fridman (08:46.840)
How do you think about that?
Kevin Scott (08:47.680)
I think it's sort of one of the most important questions
Lex Fridman (08:51.140)
in technology like maybe even in society right now
Kevin Scott (08:54.920)
about how is AI going to develop
Lex Fridman (08:59.640)
over the course of the next several decades
Lex Fridman (09:01.960)
and what's it going to be used for
Lex Fridman (09:03.600)
and what benefits will it produce
Lex Fridman (09:06.520)
and what negative impacts will it produce
Lex Fridman (09:08.480)
and who gets to steer this whole thing.
Kevin Scott (09:13.960)
I'll say at the highest level,
Lex Fridman (09:17.240)
one of the real joys of getting to do what I do at Microsoft
Kevin Scott (09:22.920)
is Microsoft has this heritage as a platform company
Lex Fridman (09:27.560)
and so Bill has this thing that he said a bunch of years ago
Kevin Scott (09:32.880)
where the measure of a successful platform
Lex Fridman (09:36.440)
is that it produces far more economic value
Kevin Scott (09:39.800)
for the people who build on top of the platform
Lex Fridman (09:41.820)
than is created for the platform owner or builder
Lex Fridman (09:47.320)
and I think we have to think about AI that way.
Lex Fridman (09:51.160)
As a platform.
Kevin Scott (09:52.240)
Yeah, it has to be a platform that other people can use
Lex Fridman (09:56.260)
to build businesses, to fulfill their creative objectives,
Kevin Scott (10:01.280)
to be entrepreneurs, to solve problems that they have
Lex Fridman (10:04.640)
in their work and in their lives.
Kevin Scott (10:07.680)
It can't be a thing where there are a handful of companies
Lex Fridman (10:11.960)
sitting in a very small handful of cities geographically
Kevin Scott (10:16.440)
who are making all the decisions about what goes into the AI
Lex Fridman (10:21.440)
and then on top of all this infrastructure,
Kevin Scott (10:26.880)
then build all of the commercially valuable uses for it.
Lex Fridman (10:30.960)
So I think that's bad from a sort of economics
Lex Fridman (10:36.480)
and sort of equitable distribution of value perspective,
Lex Fridman (10:40.120)
sort of back to this whole notion of did the markets work?
Lex Fridman (10:44.520)
But I think it's also bad from an innovation perspective
Lex Fridman (10:47.560)
because I have infinite amounts of faith
Kevin Scott (10:51.360)
in human beings that if you give folks powerful tools,
Lex Fridman (10:55.720)
they will go do interesting things
Lex Fridman (10:58.240)
and it's more than just a few tens of thousands of people
Lex Fridman (11:02.280)
with the interesting tools,
Kevin Scott (11:03.340)
it should be millions of people with the tools.
Lex Fridman (11:05.380)
So it's sort of like you think about the steam engine
Kevin Scott (11:10.160)
in the late 18th century, like it was maybe the first
Lex Fridman (11:14.480)
large scale substitute for human labor
Kevin Scott (11:16.760)
that we've built like a machine
Lex Fridman (11:19.080)
and in the beginning when these things are getting deployed,
Kevin Scott (11:23.480)
the folks who got most of the value from the steam engines
Lex Fridman (11:28.280)
were the folks who had capital
Lex Fridman (11:30.140)
so they could afford to build them
Lex Fridman (11:31.560)
and like they built factories around them and businesses
Lex Fridman (11:34.680)
and the experts who knew how to build and maintain them.
Lex Fridman (11:38.640)
But access to that technology democratized over time.
Kevin Scott (11:42.840)
Like now, like an engine, it's not like a differentiated
Lex Fridman (11:47.840)
thing, like there isn't one engine company
Kevin Scott (11:50.260)
that builds all the engines
Lex Fridman (11:51.500)
and all of the things that use engines
Kevin Scott (11:53.100)
are made by this company
Lex Fridman (11:54.220)
and like they get all the economics from all of that.
Kevin Scott (11:57.420)
Like fully demarcated, like they're probably,
Lex Fridman (12:00.540)
we're sitting here in this room
Lex Fridman (12:02.300)
and like even though they're probably things
Lex Fridman (12:05.220)
like the MEMS gyroscope that are in both of our phones,
Kevin Scott (12:09.100)
like there's like little engines sort of everywhere.
Lex Fridman (12:13.220)
They're just a component in how we build the modern world.
Kevin Scott (12:16.260)
Like AI needs to get there.
Lex Fridman (12:17.700)
Yeah, so that's a really powerful way to think.
Kevin Scott (12:20.220)
If we think of AI as a platform
Lex Fridman (12:22.700)
versus a tool that Microsoft owns,
Kevin Scott (12:26.860)
as a platform that enables creation on top of it,
Lex Fridman (12:30.140)
that's the way to democratize it.
Kevin Scott (12:31.500)
That's really interesting actually.
Lex Fridman (12:34.220)
And Microsoft throughout its history
Kevin Scott (12:36.060)
has been positioned well to do that.
Lex Fridman (12:38.260)
And the tie back to this radical markets thing,
Kevin Scott (12:41.660)
like so my team has been working with Glenn on this,
Lex Fridman (12:49.100)
and Jaren Lanier actually.
Lex Fridman (12:50.940)
So Jaren is the sort of father of virtual reality.
Lex Fridman (12:56.180)
Like he's one of the most interesting human beings on the planet,
Kevin Scott (13:00.100)
like a sweet, sweet guy.
Lex Fridman (13:02.220)
And so Jaren and Glenn and folks in my team have been working
Kevin Scott (13:07.660)
on this notion of data as labor
Lex Fridman (13:10.300)
or like they call it data dignity as well.
Lex Fridman (13:13.100)
And so the idea is that if you,
Lex Fridman (13:16.700)
again going back to this sort of industrial analogy,
Kevin Scott (13:20.580)
if you think about data as the raw material that is
Lex Fridman (13:24.700)
consumed by the machine of AI in order to do useful things,
Kevin Scott (13:30.060)
then like we're not doing a really great job right now in having
Lex Fridman (13:34.940)
transparent marketplaces for valuing those data contributions.
Lex Fridman (13:39.580)
So and we all make them explicitly like you go to LinkedIn,
Lex Fridman (13:43.540)
you sort of set up your profile on LinkedIn,
Kevin Scott (13:46.140)
like that's an explicit contribution.
Lex Fridman (13:47.780)
Like you know exactly the information
Kevin Scott (13:49.460)
that you're putting into the system.
Lex Fridman (13:50.700)
And like you put it there because you have
Kevin Scott (13:52.780)
some nominal notion of what value you're going to get in return.
Lex Fridman (13:56.620)
But it's like only nominal,
Kevin Scott (13:57.700)
like you don't know exactly what value you're getting in return.
Lex Fridman (14:00.460)
Like service is free,
Kevin Scott (14:01.860)
like it's low amount of perceived debt.
Lex Fridman (14:04.620)
And then you've got all this indirect contribution that you're
Kevin Scott (14:06.900)
making just by virtue of interacting with all of
Lex Fridman (14:09.540)
the technology that's in your daily life.
Lex Fridman (14:13.180)
And so like what Glenn and
Lex Fridman (14:15.580)
Jaren and this data dignity team are trying to do is like,
Kevin Scott (14:19.340)
can we figure out a set of mechanisms that let us value
Lex Fridman (14:23.820)
those data contributions so that you could create
Kevin Scott (14:27.260)
an economy and like a set of controls and incentives that
Lex Fridman (14:31.700)
would allow people to like maybe even in the limit,
Kevin Scott (14:36.860)
like earn part of their living
Lex Fridman (14:38.860)
through the data that they're creating.
Lex Fridman (14:41.020)
And like you can sort of see it in explicit ways.
Lex Fridman (14:42.660)
There are these companies like Scale AI,
Lex Fridman (14:46.020)
and like there are a whole bunch of them in China
Lex Fridman (14:49.420)
right now that are basically data labeling companies.
Lex Fridman (14:52.380)
So like you're doing supervised machine learning,
Lex Fridman (14:54.540)
you need lots and lots of label training data.
Lex Fridman (14:57.900)
And like those people who work for
Lex Fridman (15:01.540)
those companies are getting compensated
Kevin Scott (15:03.460)
for their data contributions into the system.
Lex Fridman (15:06.180)
And so.
Kevin Scott (15:07.380)
That's easier to put a number on
Lex Fridman (15:09.500)
their contribution because they're explicitly labeling data.
Kevin Scott (15:11.980)
Correct.
Lex Fridman (15:12.380)
But you're saying that we're all
Kevin Scott (15:13.620)
contributing data in different kinds of ways.
Lex Fridman (15:15.540)
And it's fascinating to start to
Kevin Scott (15:18.260)
explicitly try to put a number on it.
Lex Fridman (15:20.860)
Do you think that's possible?
Kevin Scott (15:22.580)
I don't know. It's hard. It really is.
Lex Fridman (15:24.980)
Because we don't have
Kevin Scott (15:29.420)
as much transparency as I think
Lex Fridman (15:33.740)
we need in like how the data is getting used.
Lex Fridman (15:37.220)
And it's super complicated.
Lex Fridman (15:38.660)
Like we, I think as
Kevin Scott (15:41.300)
technologists sort of appreciate
Lex Fridman (15:42.860)
like some of the subtlety there.
Kevin Scott (15:44.140)
It's like the data gets created and then it gets,
Lex Fridman (15:48.820)
it's not valuable.
Kevin Scott (15:50.940)
Like the data exhaust that you give off,
Lex Fridman (15:55.740)
or the explicit data that I am putting into
Kevin Scott (16:00.580)
the system isn't super valuable atomically.
Lex Fridman (16:05.100)
Like it's only valuable when you sort of
Kevin Scott (16:07.260)
aggregate it together into sort of large numbers.
Lex Fridman (16:10.420)
This is true even for these like folks who are
Kevin Scott (16:12.100)
getting compensated for like labeling things.
Lex Fridman (16:14.860)
Like for supervised machine learning now,
Kevin Scott (16:16.460)
like you need lots of labels to
Lex Fridman (16:18.860)
train a model that performs well.
Lex Fridman (16:21.900)
And so I think that's one of the challenges.
Lex Fridman (16:24.420)
It's like how do you sort of figure
Kevin Scott (16:27.220)
out like because this data is getting combined in
Lex Fridman (16:29.900)
so many ways like through
Kevin Scott (16:32.620)
these combinations like how the value is flowing.
Lex Fridman (16:35.700)
Yeah, that's fascinating.
Kevin Scott (16:37.620)
Yeah. And it's fascinating that you're thinking about this.
Lex Fridman (16:41.860)
And I wasn't even going into this conversation expecting
Kevin Scott (16:44.980)
the breadth of research really
Lex Fridman (16:48.180)
that Microsoft broadly is thinking about,
Kevin Scott (16:50.580)
you're thinking about at Microsoft.
Lex Fridman (16:52.100)
So if we go back to 89 when Microsoft released Office,
Kevin Scott (16:57.580)
or 1990 when they released Windows 3.0.
Lex Fridman (17:02.740)
In your view, I know you weren't there through its history,
Lex Fridman (17:07.980)
but how has the company changed in
Lex Fridman (17:09.940)
the 30 years since as you look at it now?
Kevin Scott (17:12.580)
The good thing is it's started off as a platform company.
Lex Fridman (17:16.900)
Like it's still a platform company,
Kevin Scott (17:20.020)
like the parts of the business that are thriving and
Lex Fridman (17:22.940)
most successful are those that are building platforms.
Kevin Scott (17:26.660)
Like the mission of the company now is,
Lex Fridman (17:28.980)
the mission's changed.
Kevin Scott (17:30.220)
It's like changed in a very interesting way.
Lex Fridman (17:32.380)
So back in 89,
Kevin Scott (17:35.860)
90 like they were still on the original mission,
Lex Fridman (17:39.100)
which was like put a PC on every desk and in every home.
Lex Fridman (17:43.940)
And it was basically about democratizing access to
Lex Fridman (17:47.700)
this new personal computing technology,
Kevin Scott (17:50.140)
which when Bill started the company,
Lex Fridman (17:52.540)
integrated circuit microprocessors were a brand new thing.
Lex Fridman (17:57.740)
And people were building homebrew computers from kits,
Lex Fridman (18:03.900)
like the way people build ham radios right now.
Kevin Scott (18:08.140)
I think this is the interesting thing
Lex Fridman (18:10.700)
for folks who build platforms in general.
Kevin Scott (18:12.860)
Bill saw the opportunity there and
Lex Fridman (18:17.060)
what personal computers could do.
Lex Fridman (18:18.780)
And it was like, it was sort of a reach.
Lex Fridman (18:20.500)
Like you just sort of imagine like where things
Kevin Scott (18:22.460)
were when they started the company
Lex Fridman (18:24.900)
versus where things are now.
Kevin Scott (18:26.100)
Like in success,
Lex Fridman (18:27.860)
when you've democratized a platform,
Kevin Scott (18:29.400)
it just sort of vanishes into the platform.
Lex Fridman (18:31.020)
You don't pay attention to it anymore.
Kevin Scott (18:32.500)
Like operating systems aren't a thing anymore.
Lex Fridman (18:35.420)
Like they're super important,
Kevin Scott (18:36.780)
like completely critical.
Lex Fridman (18:38.020)
And like when you see one fail,
Kevin Scott (18:41.460)
like you just sort of understand.
Lex Fridman (18:43.500)
But like it's not a thing where you're not like
Kevin Scott (18:46.060)
waiting for the next operating system thing
Lex Fridman (18:50.220)
in the same way that you were in 1995, right?
Kevin Scott (18:52.860)
Like in 1995, like we had
Lex Fridman (18:54.500)
Rolling Stones on the stage with the Windows 95 rollout.
Kevin Scott (18:57.580)
Like it was like the biggest thing in the world.
Lex Fridman (18:59.300)
Everybody lined up for it the way
Kevin Scott (19:01.340)
that people used to line up for iPhone.
Lex Fridman (19:03.380)
But like, you know, eventually,
Lex Fridman (19:04.820)
and like this isn't necessarily a bad thing.
Lex Fridman (19:07.160)
Like it just sort of, you know,
Kevin Scott (19:08.820)
the success is that it's sort of, it becomes ubiquitous.
Lex Fridman (19:12.860)
It's like everywhere, like human beings,
Kevin Scott (19:14.780)
when their technology becomes ubiquitous,
Lex Fridman (19:16.580)
they just sort of start taking it for granted.
Lex Fridman (19:18.180)
So the mission now that Satya
Lex Fridman (19:22.100)
rearticulated five plus years ago now,
Kevin Scott (19:25.220)
when he took over as CEO of the company.
Lex Fridman (19:28.260)
Our mission is to empower every individual and
Kevin Scott (19:33.620)
every organization in the world to be more successful.
Lex Fridman (19:38.340)
And so, you know, again,
Kevin Scott (19:40.860)
like that's a platform mission.
Lex Fridman (19:43.100)
And like the way that we do it now is, is different.
Kevin Scott (19:46.300)
It's like we have a hyperscale cloud that
Lex Fridman (19:48.780)
people are building their applications on top of.
Kevin Scott (19:51.620)
Like we have a bunch of AI infrastructure that
Lex Fridman (19:53.740)
people are building their AI applications on top of.
Kevin Scott (19:56.220)
We have, you know,
Lex Fridman (19:58.100)
we have a productivity suite of software,
Kevin Scott (20:02.060)
like Microsoft Dynamics, which, you know,
Lex Fridman (20:05.740)
some people might not think is the sexiest thing in the world,
Lex Fridman (20:07.820)
but it's like helping people figure out how to automate
Lex Fridman (20:10.820)
all of their business processes and workflows
Lex Fridman (20:13.580)
and to help those businesses using it to grow and be more.
Lex Fridman (20:19.060)
So it's a much broader vision
Kevin Scott (20:23.180)
in a way now than it was back then.
Lex Fridman (20:25.460)
Like it was sort of very particular thing.
Lex Fridman (20:27.380)
And like now, like we live in this world where
Lex Fridman (20:29.380)
technology is so powerful and it's like
Kevin Scott (20:32.380)
such a basic fact of life that it both exists
Lex Fridman (20:39.700)
and is going to get better and better over time
Kevin Scott (20:42.700)
or at least more and more powerful over time.
Lex Fridman (20:45.980)
So like, you know, what you have to do as a platform player
Kevin Scott (20:48.140)
is just much bigger.
Lex Fridman (20:49.900)
Right. There's so many directions in which you can transform.
Kevin Scott (20:52.980)
You didn't mention mixed reality, too.
Lex Fridman (20:55.140)
You know, that's probably early days
Kevin Scott (20:59.140)
or it depends how you think of it.
Lex Fridman (21:00.620)
But if we think on a scale of centuries,
Kevin Scott (21:02.140)
it's the early days of mixed reality.
Lex Fridman (21:04.020)
Oh, for sure.
Lex Fridman (21:04.900)
And so with HoloLens,
Lex Fridman (21:08.420)
Microsoft is doing some really interesting work there.
Lex Fridman (21:10.580)
Do you touch that part of the effort?
Lex Fridman (21:13.540)
What's the thinking?
Lex Fridman (21:14.820)
Do you think of mixed reality as a platform, too?
Lex Fridman (21:17.620)
Oh, sure.
Kevin Scott (21:18.460)
When we look at what the platforms of the future could be,
Lex Fridman (21:21.300)
it's like fairly obvious that like AI is one.
Kevin Scott (21:23.900)
Like you don't have to, I mean, like that's,
Lex Fridman (21:26.580)
you know, you sort of say it to like someone
Lex Fridman (21:29.140)
and you know, like they get it.
Lex Fridman (21:31.940)
But like we also think of the like mixed reality
Lex Fridman (21:36.300)
and quantum as like these two interesting,
Lex Fridman (21:39.580)
you know, potentially.
Lex Fridman (21:40.900)
Quantum computing?
Lex Fridman (21:41.820)
Yeah.
Kevin Scott (21:42.660)
Okay. So let's get crazy then.
Lex Fridman (21:44.500)
So you're talking about some futuristic things here.
Kevin Scott (21:48.900)
Well, the mixed reality, Microsoft is really,
Lex Fridman (21:50.860)
it's not even futuristic, it's here.
Kevin Scott (21:52.620)
It is.
Lex Fridman (21:53.460)
It's incredible stuff.
Lex Fridman (21:54.300)
And look, and it's having an impact right now.
Lex Fridman (21:56.660)
Like one of the more interesting things
Kevin Scott (21:58.740)
that's happened with mixed reality
Lex Fridman (21:59.980)
over the past couple of years that I didn't clearly see
Kevin Scott (22:04.140)
is that it's become the computing device
Lex Fridman (22:08.420)
for folks who, for doing their work,
Kevin Scott (22:13.180)
who haven't used any computing device at all
Lex Fridman (22:16.060)
to do their work before.
Lex Fridman (22:16.980)
So technicians and service folks and people
Lex Fridman (22:21.500)
who are doing like machine maintenance on factory floors.
Lex Fridman (22:25.340)
So like they, you know, because they're mobile
Lex Fridman (22:28.780)
and like they're out in the world
Lex Fridman (22:30.300)
and they're working with their hands
Lex Fridman (22:32.340)
and, you know, sort of servicing these like
Kevin Scott (22:34.260)
very complicated things, they're,
Lex Fridman (22:37.460)
they don't use their mobile phone
Lex Fridman (22:39.420)
and like they don't carry a laptop with them
Lex Fridman (22:41.420)
and, you know, they're not tethered to a desk.
Lex Fridman (22:43.500)
And so mixed reality, like where it's getting traction
Lex Fridman (22:47.340)
right now, where HoloLens is selling a lot of units
Kevin Scott (22:50.740)
is for these sorts of applications for these workers.
Lex Fridman (22:54.580)
And it's become like, I mean, like the people love it.
Kevin Scott (22:58.060)
They're like, oh my God, like this is like for them,
Lex Fridman (23:01.140)
like the same sort of productivity boosts that,
Kevin Scott (23:03.460)
you know, like an office worker had
Lex Fridman (23:05.500)
when they got their first personal computer.
Kevin Scott (23:08.220)
Yeah, but you did mention it's certainly obvious AI
Lex Fridman (23:12.100)
as a platform, but can we dig into it a little bit?
Lex Fridman (23:15.580)
How does AI begin to infuse some of the products
Lex Fridman (23:18.300)
in Microsoft?
Lex Fridman (23:19.500)
So currently providing training of,
Lex Fridman (23:24.500)
for example, neural networks in the cloud
Kevin Scott (23:26.700)
or providing pre trained models or just even providing
Lex Fridman (23:34.300)
computing resources and whatever different inference
Kevin Scott (23:37.540)
that you wanna do using neural networks.
Lex Fridman (23:39.940)
How do you think of AI infusing as a platform
Lex Fridman (23:44.500)
that Microsoft can provide?
Lex Fridman (23:45.900)
Yeah, I mean, I think it's super interesting.
Kevin Scott (23:48.340)
It's like everywhere.
Lex Fridman (23:49.580)
And like we run these review meetings now
Kevin Scott (23:54.580)
where it's me and Satya and like members
Lex Fridman (24:00.700)
of Satya's leadership team and like a cross functional
Kevin Scott (24:04.340)
group of folks across the entire company
Lex Fridman (24:06.180)
who are working on like either AI infrastructure
Kevin Scott (24:11.820)
or like have some substantial part of their product work
Lex Fridman (24:18.900)
using AI in some significant way.
Kevin Scott (24:22.580)
Now, the important thing to understand
Lex Fridman (24:23.940)
is like when you think about like how the AI
Kevin Scott (24:26.620)
is gonna manifest in like an experience
Lex Fridman (24:29.980)
for something that's gonna make it better,
Kevin Scott (24:31.820)
like I think you don't want the AIness
Lex Fridman (24:36.900)
to be the first order thing.
Kevin Scott (24:38.780)
It's like whatever the product is
Lex Fridman (24:40.900)
and like the thing that is trying to help you do,
Kevin Scott (24:43.700)
like the AI just sort of makes it better.
Lex Fridman (24:45.620)
And this is a gross exaggeration,
Lex Fridman (24:47.900)
but like people get super excited about like
Lex Fridman (24:51.580)
where the AI is showing up in products and I'm like,
Lex Fridman (24:54.220)
do you get that excited about like
Lex Fridman (24:55.780)
where you're using a hash table like in your code?
Kevin Scott (24:59.660)
Like it's just another.
Lex Fridman (25:01.100)
It's just a tool.
Kevin Scott (25:01.940)
It's a very interesting programming tool,
Lex Fridman (25:03.780)
but it's sort of like it's an engineering tool.
Lex Fridman (25:07.340)
And so like it shows up everywhere.
Lex Fridman (25:09.340)
So like we've got dozens and dozens of features
Kevin Scott (25:12.300)
now in Office that are powered by
Lex Fridman (25:15.660)
like fairly sophisticated machine learning,
Kevin Scott (25:18.060)
our search engine wouldn't work at all
Lex Fridman (25:21.980)
if you took the machine learning out of it.
Kevin Scott (25:24.620)
The like increasingly things like content moderation
Lex Fridman (25:30.860)
on our Xbox and xCloud platform.
Kevin Scott (25:36.820)
When you mean moderation,
Lex Fridman (25:37.900)
you mean like the recommender is like showing
Lex Fridman (25:39.500)
what you wanna look at next.
Lex Fridman (25:41.540)
No, no, no, it's like anti bullying stuff.
Lex Fridman (25:43.780)
So the usual social network stuff
Lex Fridman (25:45.780)
that you have to deal with.
Kevin Scott (25:46.820)
Yeah, correct.
Lex Fridman (25:47.660)
But it's like really it's targeted,
Kevin Scott (25:49.860)
it's targeted towards a gaming audience.
Lex Fridman (25:52.060)
So it's like a very particular type of thing
Kevin Scott (25:54.580)
where the line between playful banter
Lex Fridman (25:59.260)
and like legitimate bullying is like a subtle one.
Lex Fridman (26:02.100)
And like you have to like, it's sort of tough.
Lex Fridman (26:05.860)
Like I have.
Kevin Scott (26:07.340)
I'd love to if we could dig into it
Lex Fridman (26:08.860)
because you're also,
Kevin Scott (26:10.060)
you led the engineering efforts of LinkedIn.
Lex Fridman (26:12.980)
And if we look at LinkedIn as a social network,
Lex Fridman (26:17.460)
and if we look at the Xbox gaming as the social components,
Lex Fridman (26:21.700)
the very different kinds of I imagine communication
Lex Fridman (26:24.780)
going on on the two platforms, right?
Lex Fridman (26:26.740)
And the line in terms of bullying and so on
Kevin Scott (26:29.420)
is different on the platforms.
Lex Fridman (26:31.420)
So how do you,
Kevin Scott (26:33.140)
I mean, it's such a fascinating philosophical discussion
Lex Fridman (26:36.180)
of where that line is.
Kevin Scott (26:37.140)
I don't think anyone knows the right answer.
Lex Fridman (26:39.780)
Twitter folks are under fire now, Jack at Twitter
Kevin Scott (26:43.260)
for trying to find that line.
Lex Fridman (26:45.060)
Nobody knows what that line is.
Lex Fridman (26:46.860)
But how do you try to find the line
Lex Fridman (26:50.940)
for trying to prevent abusive behavior
Lex Fridman (26:57.940)
and at the same time, let people be playful
Lex Fridman (27:00.140)
and joke around and that kind of thing?
Kevin Scott (27:02.780)
I think in a certain way,
Lex Fridman (27:03.980)
like if you have what I would call vertical social networks,
Kevin Scott (27:10.300)
it gets to be a little bit easier.
Lex Fridman (27:12.140)
So like if you have a clear notion
Kevin Scott (27:14.380)
of like what your social network should be used for,
Lex Fridman (27:17.940)
or like what you are designing a community around,
Kevin Scott (27:22.220)
then you don't have as many dimensions
Lex Fridman (27:25.740)
to your sort of content safety problem
Kevin Scott (27:28.900)
as you do in a general purpose platform.
Lex Fridman (27:33.700)
I mean, so like on LinkedIn,
Kevin Scott (27:37.460)
like the whole social network
Lex Fridman (27:38.820)
is about connecting people with opportunity,
Kevin Scott (27:41.540)
whether it's helping them find a job
Lex Fridman (27:43.140)
or to sort of find mentors
Kevin Scott (27:46.380)
or to sort of help them like find their next sales lead
Lex Fridman (27:52.180)
or to just sort of allow them to broadcast
Kevin Scott (27:56.220)
their sort of professional identity
Lex Fridman (27:59.500)
to their network of peers and collaborators
Lex Fridman (28:06.740)
and sort of professional community.
Lex Fridman (28:08.300)
Like that is, I mean, like in some ways,
Kevin Scott (28:09.940)
like that's very, very broad,
Lex Fridman (28:11.580)
but in other ways it's sort of, it's narrow.
Lex Fridman (28:15.180)
And so like you can build AI's like machine learning systems
Lex Fridman (28:20.980)
that are capable with those boundaries
Kevin Scott (28:25.620)
of making better automated decisions
Lex Fridman (28:28.100)
about like what is sort of inappropriate
Lex Fridman (28:30.740)
and offensive comment or dangerous comment
Lex Fridman (28:32.940)
or illegal content when you have some constraints.
Kevin Scott (28:37.940)
You know, same thing with like the gaming social network.
Lex Fridman (28:43.900)
So for instance, like it's about playing games,
Kevin Scott (28:45.740)
not having fun.
Lex Fridman (28:47.260)
And like the thing that you don't want to have happen
Kevin Scott (28:49.460)
on the platform is why bullying is such an important thing.
Lex Fridman (28:52.220)
Like bullying is not fun.
Lex Fridman (28:53.740)
So you want to do everything in your power
Lex Fridman (28:56.460)
to encourage that not to happen.
Lex Fridman (28:59.380)
And yeah, but I think it's sort of a tough problem
Lex Fridman (29:03.420)
in general and it's one where I think, you know,
Kevin Scott (29:05.260)
eventually we're going to have to have some sort
Lex Fridman (29:09.980)
of clarification from our policymakers about what it is
Kevin Scott (29:15.940)
that we should be doing, like where the lines are,
Lex Fridman (29:18.940)
because it's tough.
Lex Fridman (29:20.860)
Like you don't, like in democracy, right?
Lex Fridman (29:23.740)
Like you don't want,
Kevin Scott (29:25.540)
you want some sort of democratic involvement.
Lex Fridman (29:28.900)
Like people should have a say
Kevin Scott (29:30.460)
in like where the lines are drawn.
Lex Fridman (29:34.660)
Like you don't want a bunch of people making
Kevin Scott (29:37.500)
like unilateral decisions.
Lex Fridman (29:39.460)
And like we are in a state right now
Kevin Scott (29:43.140)
for some of these platforms
Lex Fridman (29:44.220)
where you actually do have to make unilateral decisions
Kevin Scott (29:46.260)
where the policymaking isn't going to happen fast enough
Lex Fridman (29:48.620)
in order to like prevent very bad things from happening.
Lex Fridman (29:52.540)
But like we need the policymaking side of that to catch up,
Lex Fridman (29:56.020)
I think, as quickly as possible
Kevin Scott (29:58.460)
because you want that whole process to be a democratic thing,
Lex Fridman (30:01.980)
not a, you know, not some sort of weird thing
Kevin Scott (30:05.740)
where you've got a non representative group
Lex Fridman (30:08.020)
of people making decisions that have, you know,
Kevin Scott (30:10.420)
like national and global impact.
Lex Fridman (30:12.500)
And it's fascinating because the digital space is different
Kevin Scott (30:15.580)
than the physical space in which nations
Lex Fridman (30:18.340)
and governments were established.
Lex Fridman (30:19.860)
And so what policy looks like globally,
Lex Fridman (30:23.980)
what bullying looks like globally,
Kevin Scott (30:25.740)
what's healthy communication looks like globally
Lex Fridman (30:28.420)
is an open question and we're all figuring it out together,
Kevin Scott (30:31.900)
which is fascinating.
Lex Fridman (30:33.260)
Yeah, I mean with, you know, sort of fake news, for instance.
Kevin Scott (30:37.220)
And...
Lex Fridman (30:38.740)
Deep fakes and fake news generated by humans?
Kevin Scott (30:42.380)
Yeah, so we can talk about deep fakes,
Lex Fridman (30:44.660)
like I think that is another like, you know,
Kevin Scott (30:46.180)
sort of very interesting level of complexity.
Lex Fridman (30:48.340)
But like if you think about just the written word, right?
Kevin Scott (30:51.540)
Like we have, you know, we invented papyrus,
Lex Fridman (30:54.460)
what, 3,000 years ago where we, you know,
Kevin Scott (30:56.820)
you could sort of put word on paper.
Lex Fridman (31:01.220)
And then 500 years ago, like we get the printing press,
Kevin Scott (31:07.300)
like where the word gets a little bit more ubiquitous.
Lex Fridman (31:11.540)
And then like you really, really didn't get ubiquitous
Kevin Scott (31:14.660)
printed word until the end of the 19th century
Lex Fridman (31:18.500)
when the offset press was invented.
Lex Fridman (31:20.780)
And then, you know, just sort of explodes
Lex Fridman (31:22.460)
and like, you know, the cross product of that
Lex Fridman (31:25.420)
and the Industrial Revolution's need
Lex Fridman (31:28.980)
for educated citizens resulted in like
Kevin Scott (31:32.900)
this rapid expansion of literacy
Lex Fridman (31:34.780)
and the rapid expansion of the word.
Lex Fridman (31:36.060)
But like we had 3,000 years up to that point
Lex Fridman (31:39.740)
to figure out like how to, you know,
Kevin Scott (31:43.300)
like what's journalism, what's editorial integrity,
Lex Fridman (31:46.940)
like what's, you know, what's scientific peer review.
Lex Fridman (31:50.140)
And so like you built all of this mechanism
Lex Fridman (31:52.860)
to like try to filter through all of the noise
Kevin Scott (31:57.060)
that the technology made possible
Lex Fridman (31:59.820)
to like, you know, sort of getting to something
Kevin Scott (32:01.900)
that society could cope with.
Lex Fridman (32:03.980)
And like, if you think about just the piece,
Kevin Scott (32:06.580)
the PC didn't exist 50 years ago.
Lex Fridman (32:09.780)
And so in like this span of, you know,
Kevin Scott (32:11.780)
like half a century, like we've gone from no digital,
Lex Fridman (32:16.140)
you know, no ubiquitous digital technology
Kevin Scott (32:18.300)
to like having a device that sits in your pocket
Lex Fridman (32:21.060)
where you can sort of say whatever is on your mind
Kevin Scott (32:23.740)
to like what did Mary have in her,
Lex Fridman (32:27.100)
Mary Meeker just released her new like slide deck last week.
Kevin Scott (32:32.420)
You know, we've got 50% penetration of the internet
Lex Fridman (32:37.340)
to the global population.
Kevin Scott (32:38.500)
Like there are like three and a half billion people
Lex Fridman (32:40.260)
who are connected now.
Lex Fridman (32:41.740)
So it's like, it's crazy, crazy, like inconceivable,
Lex Fridman (32:44.980)
like how fast all of this happened.
Kevin Scott (32:46.460)
So, you know, it's not surprising
Lex Fridman (32:48.700)
that we haven't figured out what to do yet,
Lex Fridman (32:50.980)
but like we gotta really like lean into this set of problems
Lex Fridman (32:55.660)
because like we basically have three millennia worth of work
Kevin Scott (33:00.220)
to do about how to deal with all of this
Lex Fridman (33:02.500)
and like probably what, you know,
Kevin Scott (33:04.580)
amounts to the next decade worth of time.
Lex Fridman (33:07.020)
So since we're on the topic of tough, you know,
Kevin Scott (33:09.980)
tough challenging problems,
Lex Fridman (33:11.620)
let's look at more on the tooling side in AI
Kevin Scott (33:15.220)
that Microsoft is looking at is face recognition software.
Lex Fridman (33:18.420)
So there's a lot of powerful positive use cases
Kevin Scott (33:21.860)
for face recognition, but there's some negative ones
Lex Fridman (33:24.220)
and we're seeing those in different governments
Kevin Scott (33:27.180)
in the world.
Lex Fridman (33:28.140)
So how do you, how does Microsoft think about the use
Kevin Scott (33:30.900)
of face recognition software as a platform
Lex Fridman (33:35.740)
in governments and companies?
Lex Fridman (33:39.820)
How do we strike an ethical balance here?
Lex Fridman (33:42.300)
Yeah, I think we've articulated a clear point of view.
Lex Fridman (33:47.300)
So Brad Smith wrote a blog post last fall,
Lex Fridman (33:51.900)
I believe that sort of like outlined
Kevin Scott (33:54.180)
like very specifically what, you know,
Lex Fridman (33:57.100)
what our point of view is there.
Kevin Scott (33:59.340)
And, you know, I think we believe
Lex Fridman (34:01.060)
that there are certain uses
Kevin Scott (34:02.340)
to which face recognition should not be put.
Lex Fridman (34:04.740)
And we believe again,
Kevin Scott (34:06.060)
that there's a need for regulation there.
Lex Fridman (34:09.220)
Like the government should like really come in
Lex Fridman (34:11.940)
and say that, you know, this is where the lines are.
Lex Fridman (34:15.780)
And like, we very much wanted to like figuring out
Kevin Scott (34:18.380)
where the lines are, should be a democratic process.
Lex Fridman (34:20.380)
But in the short term, like we've drawn some lines
Kevin Scott (34:22.780)
where, you know, we push back against uses
Lex Fridman (34:26.180)
of face recognition technology, you know,
Kevin Scott (34:29.940)
like the city of San Francisco, for instance,
Lex Fridman (34:32.300)
I think has completely outlawed any government agency
Kevin Scott (34:36.580)
from using face recognition tech.
Lex Fridman (34:39.580)
And like that may prove to be a little bit overly broad.
Lex Fridman (34:44.580)
But for like certain law enforcement things,
Lex Fridman (34:48.820)
like you really, I would personally rather be overly
Kevin Scott (34:54.060)
sort of cautious in terms of restricting use of it
Lex Fridman (34:57.380)
until like we have, you know,
Kevin Scott (34:58.900)
sort of defined a reasonable, you know,
Lex Fridman (35:02.140)
democratically determined regulatory framework
Kevin Scott (35:04.860)
for like where we could and should use it.
Lex Fridman (35:08.820)
And, you know, the other thing there is like,
Kevin Scott (35:12.140)
we've got a bunch of research that we're doing
Lex Fridman (35:13.980)
and a bunch of progress that we've made on bias there.
Lex Fridman (35:18.380)
And like, there are all sorts of like weird biases
Lex Fridman (35:20.820)
that these models can have,
Kevin Scott (35:22.980)
like all the way from like the most noteworthy one
Lex Fridman (35:25.580)
where, you know, you may have underrepresented minorities
Kevin Scott (35:31.660)
who are like underrepresented in the training data
Lex Fridman (35:34.660)
and then you start learning like strange things.
Lex Fridman (35:39.180)
But like there are even, you know, other weird things.
Lex Fridman (35:42.100)
Like we've, I think we've seen in the public research,
Kevin Scott (35:46.460)
like models can learn strange things,
Lex Fridman (35:49.460)
like all doctors are men, for instance, just, yeah.
Kevin Scott (35:54.700)
I mean, and so like, it really is a thing
Lex Fridman (35:58.900)
where it's very important for everybody
Kevin Scott (36:03.580)
who is working on these things before they push publish,
Lex Fridman (36:08.420)
they launch the experiment, they, you know, push the code
Kevin Scott (36:12.780)
to, you know, online, or they even publish the paper
Lex Fridman (36:17.100)
that they are at least starting to think about
Lex Fridman (36:20.420)
what some of the potential negative consequences are,
Lex Fridman (36:25.260)
some of this stuff.
Kevin Scott (36:26.100)
I mean, this is where, you know, like the deep fake stuff
Lex Fridman (36:29.020)
I find very worrisome just because
Kevin Scott (36:32.340)
there are going to be some very good beneficial uses
Lex Fridman (36:39.780)
of like GAN generated imagery.
Lex Fridman (36:46.100)
And funny enough, like one of the places
Lex Fridman (36:48.460)
where it's actually useful is we're using the technology
Kevin Scott (36:52.940)
right now to generate synthetic visual data
Lex Fridman (36:58.620)
for training some of the face recognition models
Kevin Scott (37:01.140)
to get rid of the bias.
Lex Fridman (37:03.420)
So like, that's one like super good use of the tech,
Lex Fridman (37:05.740)
but like, you know, it's getting good enough now
Lex Fridman (37:09.620)
where, you know, it's going to sort of challenge
Kevin Scott (37:12.300)
a normal human being's ability to,
Lex Fridman (37:14.300)
like now you're just sort of say,
Kevin Scott (37:15.740)
like it's very expensive for someone
Lex Fridman (37:19.300)
to fabricate a photorealistic fake video.
Lex Fridman (37:24.140)
And like GANs are going to make it fantastically cheap
Lex Fridman (37:26.900)
to fabricate a photorealistic fake video.
Lex Fridman (37:30.420)
And so like what you assume you can sort of trust is true
Lex Fridman (37:34.460)
versus like be skeptical about is about to change.
Lex Fridman (37:38.380)
And like, we're not ready for it, I don't think.
Lex Fridman (37:40.540)
The nature of truth, right.
Kevin Scott (37:41.980)
That's, it's also exciting because I think both you and I
Lex Fridman (37:46.580)
probably would agree that the way to solve,
Lex Fridman (37:49.580)
to take on that challenge is with technology, right?
Lex Fridman (37:52.820)
There's probably going to be ideas of ways to verify
Kevin Scott (37:56.820)
which kind of video is legitimate, which kind is not.
Lex Fridman (38:00.820)
So to me, that's an exciting possibility,
Kevin Scott (38:03.860)
most likely for just the comedic genius
Lex Fridman (38:07.180)
that the internet usually creates with these kinds of videos
Lex Fridman (38:10.980)
and hopefully will not result in any serious harm.
Lex Fridman (38:13.940)
Yeah, and it could be, you know,
Kevin Scott (38:17.100)
like I think we will have technology to,
Lex Fridman (38:21.180)
that may be able to detect whether or not
Kevin Scott (38:23.580)
something's fake or real.
Lex Fridman (38:24.460)
Although the fakes are pretty convincing,
Kevin Scott (38:30.180)
even like when you subject them to machine scrutiny.
Lex Fridman (38:34.340)
But, you know, we also have these increasingly
Kevin Scott (38:37.820)
interesting social networks, you know,
Lex Fridman (38:40.540)
that are under fire right now
Kevin Scott (38:43.580)
for some of the bad things that they do.
Lex Fridman (38:46.220)
Like one of the things you could choose to do
Kevin Scott (38:47.700)
with a social network is like you could,
Lex Fridman (38:51.780)
you could use crypto and the networks
Kevin Scott (38:55.580)
to like have content signed
Lex Fridman (38:57.740)
where you could have a like full chain of custody
Kevin Scott (39:01.420)
that accompanied every piece of content.
Lex Fridman (39:03.900)
So like when you're viewing something
Lex Fridman (39:06.780)
and like you want to ask yourself,
Lex Fridman (39:08.540)
like how much can I trust this?
Kevin Scott (39:11.020)
Like you can click something
Lex Fridman (39:12.380)
and like have a verified chain of custody
Kevin Scott (39:14.980)
that shows like, oh, this is coming from this source.
Lex Fridman (39:19.060)
And it's like signed by like someone
Kevin Scott (39:21.660)
whose identity I trust.
Lex Fridman (39:24.100)
Yeah, I think having that, you know,
Kevin Scott (39:25.420)
having that chain of custody,
Lex Fridman (39:26.620)
like being able to like say, oh, here's this video.
Kevin Scott (39:29.340)
Like it may or may not have been produced
Lex Fridman (39:31.940)
using some of this deepfake technology,
Lex Fridman (39:33.740)
but if you've got a verified chain of custody
Lex Fridman (39:35.660)
where you can sort of trace it all the way back
Kevin Scott (39:37.780)
to an identity and you can decide whether or not
Lex Fridman (39:39.940)
like I trust this identity.
Kevin Scott (39:41.540)
Like, oh no, this is really from the White House
Lex Fridman (39:43.340)
or like this is really from the, you know,
Kevin Scott (39:45.500)
the office of this particular presidential candidate
Lex Fridman (39:48.820)
or it's really from, you know, Jeff Wiener, CEO of LinkedIn
Kevin Scott (39:53.540)
or Satya Nadella, CEO of Microsoft.
Lex Fridman (39:55.540)
Like that might be like one way
Kevin Scott (39:58.420)
that you can solve some of the problems.
Lex Fridman (39:59.940)
So like that's not the super high tech.
Kevin Scott (40:01.780)
Like we've had all of this technology forever.
Lex Fridman (40:04.500)
And, but I think you're right.
Kevin Scott (40:06.700)
Like it has to be some sort of technological thing
Lex Fridman (40:11.100)
because the underlying tech that is used to create this
Kevin Scott (40:15.820)
is not going to do anything but get better over time
Lex Fridman (40:18.780)
and the genie is sort of out of the bottle.
Kevin Scott (40:21.140)
There's no stuffing it back in.
Lex Fridman (40:22.780)
And there's a social component,
Kevin Scott (40:24.500)
which I think is really healthy for a democracy
Lex Fridman (40:26.620)
where people will be skeptical
Kevin Scott (40:28.500)
about the thing they watch in general.
Lex Fridman (40:32.140)
So, you know, which is good.
Kevin Scott (40:34.180)
Skepticism in general is good for content.
Lex Fridman (40:37.300)
So deepfakes in that sense are creating a global skepticism
Kevin Scott (40:41.780)
about can they trust what they read.
Lex Fridman (40:44.780)
It encourages further research.
Kevin Scott (40:46.900)
I come from the Soviet Union
Lex Fridman (40:49.860)
where basically nobody trusted the media
Kevin Scott (40:53.380)
because you knew it was propaganda.
Lex Fridman (40:55.180)
And that kind of skepticism encouraged further research
Kevin Scott (40:59.220)
about ideas as opposed to just trusting any one source.
Lex Fridman (41:02.420)
Well, look, I think it's one of the reasons why
Kevin Scott (41:04.340)
the scientific method and our apparatus
Lex Fridman (41:09.500)
of modern science is so good.
Kevin Scott (41:11.540)
Like, because you don't have to trust anything.
Lex Fridman (41:15.420)
Like, the whole notion of modern science
Kevin Scott (41:20.180)
beyond the fact that this is a hypothesis
Lex Fridman (41:22.460)
and this is an experiment to test the hypothesis
Lex Fridman (41:24.900)
and this is a peer review process
Lex Fridman (41:27.380)
for scrutinizing published results.
Lex Fridman (41:30.140)
But stuff's also supposed to be reproducible.
Lex Fridman (41:33.300)
So you know it's been vetted by this process,
Lex Fridman (41:35.260)
but you also are expected to publish enough detail
Lex Fridman (41:38.060)
where if you are sufficiently skeptical of the thing,
Kevin Scott (41:42.100)
you can go try to reproduce it yourself.
Lex Fridman (41:44.740)
And like, I don't know what it is.
Kevin Scott (41:47.580)
Like, I think a lot of engineers are like this
Lex Fridman (41:49.980)
where like, you know, sort of this,
Kevin Scott (41:51.940)
like your brain is sort of wired for skepticism.
Lex Fridman (41:55.580)
Like, you don't just first order trust everything
Kevin Scott (41:58.060)
that you see and encounter.
Lex Fridman (42:00.100)
And like, you're sort of curious to understand,
Kevin Scott (42:02.620)
you know, the next thing.
Lex Fridman (42:04.540)
But like, I think it's an entirely healthy thing.
Lex Fridman (42:09.140)
And like, we need a little bit more of that right now.
Lex Fridman (42:12.340)
So I'm not a large business owner.
Lex Fridman (42:16.300)
So I'm just a huge fan of many of Microsoft products.
Lex Fridman (42:23.300)
I mean, I still, actually in terms of,
Kevin Scott (42:25.460)
I generate a lot of graphics and images
Lex Fridman (42:27.060)
and I still use PowerPoint to do that.
Kevin Scott (42:28.740)
It beats Illustrator for me.
Lex Fridman (42:30.500)
Even professional sort of, it's fascinating.
Lex Fridman (42:34.540)
So I wonder, what is the future of,
Lex Fridman (42:38.460)
let's say Windows and Office look like?
Lex Fridman (42:42.020)
Is, do you see it?
Lex Fridman (42:43.940)
I mean, I remember looking forward to XP.
Lex Fridman (42:45.940)
Was it exciting when XP was released?
Lex Fridman (42:48.260)
Just like you said, I don't remember when 95 was released.
Lex Fridman (42:51.180)
But XP for me was a big celebration.
Lex Fridman (42:53.900)
And when 10 came out, I was like, oh, okay.
Kevin Scott (42:56.420)
Well, it's nice.
Lex Fridman (42:57.260)
It's a nice improvement.
Lex Fridman (42:59.100)
So what do you see the future of these products?
Lex Fridman (43:03.380)
I think there's a bunch of excite.
Kevin Scott (43:04.700)
I mean, on the Office front,
Lex Fridman (43:07.260)
there's gonna be this like increasing productivity wins
Kevin Scott (43:13.900)
that are coming out of some of these AI powered features
Lex Fridman (43:17.200)
that are coming.
Kevin Scott (43:18.040)
Like the products will sort of get smarter and smarter
Lex Fridman (43:20.000)
in like a very subtle way.
Kevin Scott (43:21.240)
Like there's not gonna be this big bang moment
Lex Fridman (43:24.260)
where like Clippy is gonna reemerge and it's gonna be.
Kevin Scott (43:28.020)
Wait a minute.
Lex Fridman (43:28.860)
Okay, we'll have to wait, wait, wait.
Lex Fridman (43:30.660)
Is Clippy coming back?
Lex Fridman (43:32.580)
But quite seriously, so injection of AI.
Kevin Scott (43:37.140)
There's not much, or at least I'm not familiar,
Lex Fridman (43:39.220)
sort of assistive type of stuff going on
Kevin Scott (43:41.340)
inside the Office products.
Lex Fridman (43:43.700)
Like a Clippy style assistant, personal assistant.
Lex Fridman (43:47.740)
Do you think that there's a possibility
Lex Fridman (43:50.740)
of that in the future?
Lex Fridman (43:52.100)
So I think there are a bunch of like very small ways
Lex Fridman (43:54.820)
in which like machine learning powered assistive things
Kevin Scott (43:58.540)
are in the product right now.
Lex Fridman (44:00.260)
So there are a bunch of interesting things.
Kevin Scott (44:04.940)
Like the auto response stuff's getting better and better.
Lex Fridman (44:09.500)
And it's like getting to the point
Kevin Scott (44:11.180)
where it can auto respond with like,
Lex Fridman (44:14.820)
okay, this person's clearly trying to schedule a meeting.
Lex Fridman (44:19.260)
So it looks at your calendar and it automatically
Lex Fridman (44:21.700)
like tries to find like a time and a space
Kevin Scott (44:24.260)
that's mutually interesting.
Lex Fridman (44:27.420)
Like we have this notion of Microsoft search
Kevin Scott (44:32.420)
at a Microsoft search where it's like not just web search,
Lex Fridman (44:34.940)
but it's like search across like all of your information
Kevin Scott (44:38.180)
that's sitting inside of like your Office 365 tenant
Lex Fridman (44:43.300)
and like potentially in other products.
Lex Fridman (44:46.900)
And like we have this thing called the Microsoft Graph
Lex Fridman (44:49.680)
that is basically an API federator that sort of like
Kevin Scott (44:53.980)
gets you hooked up across the entire breadth
Lex Fridman (44:57.980)
of like all of the, like what were information silos
Kevin Scott (45:01.640)
before they got woven together with the graph.
Lex Fridman (45:05.660)
Like that is like getting increasing,
Kevin Scott (45:07.860)
with increasing effectiveness,
Lex Fridman (45:09.140)
sort of plumbed into some of these auto response things
Kevin Scott (45:13.120)
where you're gonna be able to see the system
Lex Fridman (45:15.860)
like automatically retrieve information for you.
Kevin Scott (45:18.220)
Like if, you know, like I frequently send out,
Lex Fridman (45:21.140)
you know, emails to folks where like I can't find a paper
Kevin Scott (45:24.060)
or a document or whatnot.
Lex Fridman (45:25.380)
There's no reason why the system
Kevin Scott (45:26.340)
won't be able to do that for you.
Lex Fridman (45:27.540)
And like, I think the, it's building towards
Kevin Scott (45:31.980)
like having things that look more like,
Lex Fridman (45:34.480)
like a fully integrated, you know, assistant,
Lex Fridman (45:37.900)
but like you'll have a bunch of steps
Lex Fridman (45:40.740)
that you will see before you,
Kevin Scott (45:42.820)
like it will not be this like big bang thing
Lex Fridman (45:45.140)
where like Clippy comes back and you've got this like,
Kevin Scott (45:47.420)
you know, manifestation of, you know,
Lex Fridman (45:49.400)
like a fully, fully powered assistant.
Lex Fridman (45:53.380)
So I think that's, that's definitely coming in,
Lex Fridman (45:56.940)
like all of the, you know, collaboration,
Kevin Scott (45:58.700)
coauthoring stuff's getting better.
Lex Fridman (46:00.740)
You know, it's like really interesting.
Kevin Scott (46:02.220)
Like if you look at how we use
Lex Fridman (46:06.500)
the Office product portfolio at Microsoft,
Kevin Scott (46:09.020)
like more and more of it is happening inside of
Lex Fridman (46:12.140)
like Teams as a canvas.
Lex Fridman (46:14.500)
And like, it's this thing where, you know,
Lex Fridman (46:17.180)
you've got collaboration is like at the center
Kevin Scott (46:20.620)
of the product and like we built some like really cool stuff
Lex Fridman (46:25.620)
that's some of, which is about to be open source
Kevin Scott (46:28.420)
that are sort of framework level things
Lex Fridman (46:30.980)
for doing, for doing coauthoring.
Kevin Scott (46:34.540)
That's awesome.
Lex Fridman (46:35.380)
So in, is there a cloud component to that?
Lex Fridman (46:37.860)
So on the web, or is it,
Lex Fridman (46:40.300)
and forgive me if I don't already know this,
Lex Fridman (46:42.660)
but with Office 365, we still,
Lex Fridman (46:45.580)
the collaboration we do if we're doing Word,
Kevin Scott (46:47.540)
we still send the file around.
Lex Fridman (46:49.660)
No, no.
Lex Fridman (46:50.500)
So this is.
Lex Fridman (46:51.340)
We're already a little bit better than that.
Kevin Scott (46:54.300)
A little bit better than that and like, you know,
Lex Fridman (46:55.900)
so like the fact that you're unaware of it means
Kevin Scott (46:57.700)
we've got a better job to do,
Lex Fridman (46:59.180)
like helping you discover, discover this stuff.
Lex Fridman (47:02.900)
But yeah, I mean, it's already like got a huge,
Lex Fridman (47:06.380)
huge cloud component.
Lex Fridman (47:07.220)
And like part of, you know, part of this framework stuff,
Lex Fridman (47:09.700)
I think we're calling it, like I,
Kevin Scott (47:12.660)
like we've been working on it for a couple of years.
Lex Fridman (47:14.540)
So like, I know the internal code name for it,
Lex Fridman (47:17.220)
but I think when we launched it to build,
Lex Fridman (47:18.660)
it's called the Fluid Framework.
Kevin Scott (47:20.760)
And, but like what Fluid lets you do is like,
Lex Fridman (47:25.060)
you can go into a conversation that you're having in Teams
Lex Fridman (47:27.900)
and like reference like part of a spreadsheet
Lex Fridman (47:30.240)
that you're working on where somebody's like sitting
Kevin Scott (47:33.900)
in the Excel canvas,
Lex Fridman (47:35.580)
like working on the spreadsheet with a, you know,
Kevin Scott (47:37.740)
chart or whatnot,
Lex Fridman (47:38.860)
and like you can sort of embed like part of the spreadsheet
Kevin Scott (47:41.940)
in the Teams conversation where like you can dynamically
Lex Fridman (47:45.400)
update it and like all of the changes that you're making
Kevin Scott (47:48.740)
to the, to this object are like, you know,
Lex Fridman (47:51.220)
coordinate and everything is sort of updating in real time.
Lex Fridman (47:54.620)
So like you can be in whatever canvas is most convenient
Lex Fridman (47:57.940)
for you to get your work done.
Lex Fridman (48:00.380)
So I, out of my own sort of curiosity as an engineer,
Lex Fridman (48:03.380)
I know what it's like to sort of lead a team
Kevin Scott (48:06.220)
of 10, 15 engineers.
Lex Fridman (48:08.220)
Microsoft has, I don't know what the numbers are,
Kevin Scott (48:11.660)
maybe 50, maybe 60,000 engineers, maybe 40.
Lex Fridman (48:14.940)
I don't know exactly what the number is, it's a lot.
Kevin Scott (48:17.220)
It's tens of thousands.
Lex Fridman (48:18.900)
Right, so it's more than 10 or 15.
Kevin Scott (48:20.700)
What, I mean, you've led different sizes,
Lex Fridman (48:28.700)
mostly large size of engineers.
Lex Fridman (48:30.540)
What does it take to lead such a large group
Lex Fridman (48:33.820)
into a continue innovation,
Kevin Scott (48:37.500)
continue being highly productive
Lex Fridman (48:40.260)
and yet develop all kinds of new ideas and yet maintain,
Kevin Scott (48:44.220)
like what does it take to lead such a large group
Lex Fridman (48:47.100)
of brilliant people?
Kevin Scott (48:48.980)
I think the thing that you learn
Lex Fridman (48:52.060)
as you manage larger and larger scale
Kevin Scott (48:55.140)
is that there are three things
Lex Fridman (48:57.940)
that are like very, very important
Kevin Scott (49:00.500)
for big engineering teams.
Lex Fridman (49:02.340)
Like one is like having some sort of forethought
Kevin Scott (49:06.300)
about what it is that you're gonna be building
Lex Fridman (49:09.860)
over large periods of time.
Kevin Scott (49:11.060)
Like not exactly, like you don't need to know
Lex Fridman (49:13.100)
that like, you know, I'm putting all my chips
Kevin Scott (49:15.300)
on this one product and like this is gonna be the thing,
Lex Fridman (49:17.820)
but like it's useful to know like what sort of capabilities
Kevin Scott (49:21.460)
you think you're going to need to have
Lex Fridman (49:23.140)
to build the products of the future.
Lex Fridman (49:24.740)
And then like invest in that infrastructure,
Lex Fridman (49:28.060)
like whether, and like I'm not just talking
Kevin Scott (49:30.180)
about storage systems or cloud APIs,
Lex Fridman (49:32.740)
it's also like what does your development process look like?
Lex Fridman (49:35.380)
What tools do you want?
Lex Fridman (49:36.780)
Like what culture do you want to build around?
Kevin Scott (49:40.020)
Like how you're, you know, sort of collaborating together
Lex Fridman (49:42.780)
to like make complicated technical things.
Lex Fridman (49:45.780)
And so like having an opinion and investing in that
Lex Fridman (49:48.100)
is like, it just gets more and more important.
Lex Fridman (49:50.500)
And like the sooner you can get a concrete set of opinions,
Lex Fridman (49:54.540)
like the better you're going to be.
Kevin Scott (49:57.700)
Like you can wing it for a while at small scales,
Lex Fridman (50:01.620)
like, you know, when you start a company,
Kevin Scott (50:03.180)
like you don't have to be like super specific about it,
Lex Fridman (50:06.340)
but like the biggest miseries that I've ever seen
Kevin Scott (50:09.980)
as an engineering leader are in places
Lex Fridman (50:12.660)
where you didn't have a clear enough opinion
Kevin Scott (50:14.500)
about those things soon enough.
Lex Fridman (50:16.780)
And then you just sort of go create a bunch
Kevin Scott (50:18.740)
of technical debt and like culture debt
Lex Fridman (50:21.940)
that is excruciatingly painful to clean up.
Lex Fridman (50:25.820)
So like, that's one bundle of things.
Lex Fridman (50:28.700)
Like the other, you know, another bundle of things
Kevin Scott (50:33.260)
is like, it's just really, really important
Lex Fridman (50:36.620)
to like have a clear mission
Kevin Scott (50:41.620)
that's not just some cute crap you say
Lex Fridman (50:46.260)
because like you think you should have a mission,
Lex Fridman (50:48.940)
but like something that clarifies for people
Lex Fridman (50:52.940)
like where it is that you're headed together.
Kevin Scott (50:57.220)
Like, I know it's like probably like a little bit
Lex Fridman (50:59.180)
too popular right now,
Lex Fridman (51:00.380)
but Yuval Harari's book, Sapiens,
Lex Fridman (51:05.380)
one of the central ideas in his book is that
Kevin Scott (51:10.380)
like storytelling is like the quintessential thing
Lex Fridman (51:15.380)
for coordinating the activities of large groups of people.
Kevin Scott (51:18.780)
Like once you get past Dunbar's number,
Lex Fridman (51:21.380)
and like I've really, really seen that
Kevin Scott (51:23.980)
just managing engineering teams.
Lex Fridman (51:25.580)
Like you can just brute force things
Kevin Scott (51:30.580)
when you're less than 120, 150 folks
Lex Fridman (51:33.500)
where you can sort of know and trust
Lex Fridman (51:35.980)
and understand what the dynamics are
Lex Fridman (51:38.380)
between all the people, but like past that,
Kevin Scott (51:40.380)
like things just sort of start to catastrophically fail
Lex Fridman (51:43.780)
if you don't have some sort of set of shared goals
Kevin Scott (51:47.180)
that you're marching towards.
Lex Fridman (51:48.980)
And so like, even though it sounds touchy feely
Lex Fridman (51:51.380)
and you know, like a bunch of technical people
Lex Fridman (51:54.180)
will sort of balk at the idea that like,
Kevin Scott (51:56.180)
you need to like have a clear, like the missions,
Lex Fridman (52:00.180)
like very, very, very important.
Lex Fridman (52:02.180)
You're always right, right?
Lex Fridman (52:04.180)
Stories, that's how our society,
Kevin Scott (52:06.180)
that's the fabric that connects us,
Lex Fridman (52:08.180)
all of us is these powerful stories.
Lex Fridman (52:10.180)
And that works for companies too, right?
Lex Fridman (52:12.180)
It works for everything.
Kevin Scott (52:14.180)
Like, I mean, even down to like, you know,
Lex Fridman (52:16.180)
you sort of really think about it,
Kevin Scott (52:18.180)
like our currency, for instance, is a story.
Lex Fridman (52:20.180)
Our constitution is a story.
Kevin Scott (52:22.180)
Our laws are stories.
Lex Fridman (52:24.180)
I mean, like we believe very, very, very strongly in them.
Lex Fridman (52:27.180)
And thank God we do.
Lex Fridman (52:29.180)
But like they are,
Kevin Scott (52:31.180)
they're just abstract things.
Lex Fridman (52:33.180)
Like they're just words.
Kevin Scott (52:34.180)
Like if we don't believe in them, they're nothing.
Lex Fridman (52:36.180)
And in some sense, those stories are platforms
Lex Fridman (52:39.180)
and the kinds, some of which Microsoft is creating, right?
Lex Fridman (52:43.180)
They have platforms on which we define the future.
Lex Fridman (52:46.180)
So last question, what do you,
Lex Fridman (52:48.180)
let's get philosophical maybe,
Kevin Scott (52:50.180)
bigger than even Microsoft,
Lex Fridman (52:51.180)
what do you think the next 20, 30 plus years
Lex Fridman (52:56.180)
looks like for computing, for technology, for devices?
Lex Fridman (53:00.180)
Do you have crazy ideas about the future of the world?
Kevin Scott (53:04.180)
Yeah, look, I think we, you know,
Lex Fridman (53:06.180)
we're entering this time where we've got,
Kevin Scott (53:10.180)
we have technology that is progressing
Lex Fridman (53:13.180)
at the fastest rate that it ever has.
Lex Fridman (53:15.180)
And you've got,
Lex Fridman (53:18.180)
you've got some really big social problems,
Kevin Scott (53:21.180)
like society scale problems that we have to tackle.
Lex Fridman (53:26.180)
And so, you know, I think we're going to rise to the challenge
Lex Fridman (53:28.180)
and like figure out how to intersect
Lex Fridman (53:30.180)
like all of the power of this technology
Kevin Scott (53:32.180)
with all of the big challenges that are facing us,
Lex Fridman (53:35.180)
whether it's, you know, global warming,
Kevin Scott (53:37.180)
whether it's like the biggest remainder of the population boom
Lex Fridman (53:41.180)
is in Africa for the next 50 years or so.
Lex Fridman (53:46.180)
And like global warming is going to make it increasingly difficult
Lex Fridman (53:49.180)
to feed the global population in particular,
Kevin Scott (53:52.180)
like in this place where you're going to have
Lex Fridman (53:54.180)
like the biggest population boom.
Kevin Scott (53:57.180)
I think we, you know, like AI is going to,
Lex Fridman (54:01.180)
like if we push it in the right direction,
Kevin Scott (54:03.180)
like it can do like incredible things to empower all of us
Lex Fridman (54:07.180)
to achieve our full potential and to, you know,
Kevin Scott (54:12.180)
like live better lives.
Lex Fridman (54:15.180)
But like that also means focus on like
Kevin Scott (54:20.180)
some super important things.
Lex Fridman (54:21.180)
Like how can you apply it to healthcare to make sure that,
Kevin Scott (54:26.180)
you know, like our quality and cost of
Lex Fridman (54:29.180)
and sort of ubiquity of health coverage is better
Lex Fridman (54:33.180)
and better over time.
Lex Fridman (54:35.180)
Like that's more and more important every day is like
Kevin Scott (54:38.180)
in the United States and like the rest of the industrialized world,
Lex Fridman (54:43.180)
so Western Europe, China, Japan, Korea,
Kevin Scott (54:45.180)
like you've got this population bubble of like aging,
Lex Fridman (54:50.180)
working, you know, working age folks who are,
Kevin Scott (54:54.180)
you know, at some point over the next 20, 30 years,
Lex Fridman (54:56.180)
they're going to be largely retired.
Lex Fridman (54:58.180)
And like you're going to have more retired people
Lex Fridman (55:00.180)
than working age people.
Lex Fridman (55:01.180)
And then like you've got, you know,
Lex Fridman (55:02.180)
sort of natural questions about who's going to take care of
Kevin Scott (55:05.180)
all the old folks and who's going to do all the work.
Lex Fridman (55:07.180)
And the answers to like all of these sorts of questions,
Kevin Scott (55:11.180)
like where you're sort of running into, you know,
Lex Fridman (55:13.180)
like constraints of the, you know,
Kevin Scott (55:16.180)
the world and of society has always been like
Lex Fridman (55:20.180)
what tech is going to like help us get around this?
Kevin Scott (55:23.180)
Like when I was a kid in the 70s and 80s,
Lex Fridman (55:26.180)
like we talked all the time about like population boom,
Kevin Scott (55:29.180)
population boom, like we're going to,
Lex Fridman (55:31.180)
like we're not going to be able to like feed the planet.
Lex Fridman (55:34.180)
And like we were like right in the middle of the Green Revolution
Lex Fridman (55:38.180)
where like this massive technology driven increase
Kevin Scott (55:44.180)
in crop productivity like worldwide.
Lex Fridman (55:47.180)
And like some of that was like taking some of the things
Kevin Scott (55:49.180)
that we knew in the West and like getting them distributed
Lex Fridman (55:52.180)
to the, you know, to the developing world.
Lex Fridman (55:55.180)
And like part of it were things like, you know,
Lex Fridman (55:59.180)
just smarter biology like helping us increase.
Lex Fridman (56:03.180)
And like we don't talk about like overpopulation anymore
Lex Fridman (56:08.180)
because like we can more or less,
Kevin Scott (56:10.180)
we sort of figured out how to feed the world.
Lex Fridman (56:12.180)
Like that's a technology story.
Lex Fridman (56:14.180)
And so like I'm super, super hopeful about the future
Lex Fridman (56:19.180)
and in the ways where we will be able to apply technology
Kevin Scott (56:24.180)
to solve some of these super challenging problems.
Lex Fridman (56:28.180)
Like I've, like one of the things that I'm trying to spend
Kevin Scott (56:33.180)
my time doing right now is trying to get everybody else
Lex Fridman (56:36.180)
to be hopeful as well because, you know, back to Harare,
Kevin Scott (56:39.180)
like we are the stories that we tell.
Lex Fridman (56:41.180)
Like if we, you know, if we get overly pessimistic right now
Kevin Scott (56:44.180)
about like the potential future of technology,
Lex Fridman (56:48.180)
like we, you know, like we may fail to get all of the things
Kevin Scott (56:53.180)
in place that we need to like have our best possible future.
Lex Fridman (56:56.180)
And that kind of hopeful optimism, I'm glad that you have it
Kevin Scott (57:00.180)
because you're leading large groups of engineers
Lex Fridman (57:03.180)
that are actually defining, that are writing that story,
Kevin Scott (57:06.180)
that are helping build that future, which is super exciting.
Lex Fridman (57:09.180)
And I agree with everything you said except I do hope
Kevin Scott (57:13.180)
Clippy comes back.
Lex Fridman (57:15.180)
We miss him. I speak for the people.
Kevin Scott (57:19.180)
So, Galen, thank you so much for talking to me.
Lex Fridman (57:21.180)
Thank you so much for having me. It was a pleasure.
🔗 相关节目