Unpacking how the world is using AI | Microsoft's AI Diffusion Report

0:00 A lot of times people think

0:01 about software developers or even computer scientists,

0:03 as people that need to know a programming language.

0:06 And I think that that is the wrong approach.

0:09 Like, a software developer or someone that knows how to code

0:12 is someone that can actually communicate computers to automate process.

0:17 Whether that's in English or Python doesn't matter.

0:19 So, I think it’s the wrong approach to think it's like, “Oh,

0:23 we will need less software developers because now you can code in English.” No,

0:26 no, you will have more software developers.

0:28 It’s going to become easier to code.

0:30 It's already easier to code.

0:32 That’s Juan Lavista Ferres,

0:34 Chief Data Scientist and Director of Microsoft's AI for Good Lab.

0:39 Juan’s team just dropped the latest Global AI Diffusion Report,

0:43 our real-time look at how AI is spreading and scaling around the world.

0:49 We talk about the recent surge of adoption in Asia,

0:52 how AI is changing software development, and the role trust plays in adoption.

0:58 The report shows real momentum, but also some troubling divides.

1:04 AI diffusion with Juan Lavista Ferres, up next on Tools and Weapons.

1:10 Juan, welcome back.

1:11 I think it's an exciting time to sit down and talk,

1:14 the AI for Good Lab has now published two more reports on AI diffusion.

1:19 You're starting to put these out every quarter.

1:21 Yes.

1:22 The global report shows accelerating growth for AI.

1:25 It shows a surge in Asia.

1:28 It shows how AI for coding is impacting lots of different things.

1:33 We'll get into all of that.

1:35 And your new report, for the first time,

1:37 a county-by-county usage report for the entire United States shows that we

1:43 have our own AI divide in the country between urban and rural counties.

1:48 Before we tease that apart, can you say a little bit about how you

1:54 do these calculations across the country, around the world.

1:58 What gives the AI For Good Lab at Microsoft an ability

2:03 that, perhaps is unique to prepare these kinds of estimates?

2:07 Yeah, that’s a great question.

2:08 And, Brad, thank you for the invitation here.

2:11 We have, within Microsoft, we have billions of devices out there,

2:15 and we- we can collect in a very highly privacy-preserving way, very anonymous.

2:20 We have data on how people are using, how people are using these services.

2:25 And we can report based on the most common AI models by country,

2:31 by county, how many people are using.

2:34 And then of course we are, these are not the only devices.

2:37 So, we need to control for the internet penetration on that county.

2:42 We need to control for mobile versus PC.

2:46 Once we adjust on all of this information we can report.

2:49 And this is—the great thing about this is that allows us to do

2:51 an apples to apples comparison with countries and also a trend over time.

2:56 So, the headline is: at the end of the first quarter, the end of March,

3:02 on a global basis, 17.8% of the world's people,

3:06 or world's working-age population, was using AI.

3:11 What's the definition of working-age population?

3:14 This is the—we use the definition of the OECD,

3:16 that is this—between 15 and 65 years.

3:19 15 and 65?

3:21 Yes.

3:21 Yes.

3:22 This is like a—it's a good approximation of people that are using these.

3:25 Majority of the people that we see using these PCs are within that range.

3:29 And that was a point and a half over December.

3:31 Yes.

3:32 If it remains that way, we'd see 6% growth over the course of this year.

3:38 I think in some ways what was even more

3:40 interesting is there's now 26 countries that have top 30%.

3:44 So, we're seeing this steadily grow in a great many countries.

3:49 And we always talk about who's the leader in the world?

3:52 You want to share that?

3:53 Yeah.

3:54 So, the leader remains- remains the UAE.

3:57 It's the first actually,

3:58 the first country that tops the 70% of the working age population using.

4:03 And what is amazing about the UAE is that, like,

4:05 by then we would have expected that there was like

4:08 some stopping growth given the majority of people are using it.

4:12 And UAE remains not only the top one, but one of the ones that has the biggest

4:17 growth like for the last six, six months.

4:21 So, it has been incredible.

4:22 One of the things I found interesting is how many people,

4:26 especially in governments and, you know,

4:28 in the tech sector and the press are really following this report.

4:33 Since the last report came I've been in nine countries.

4:37 And, you know, people talk about their number.

4:39 Or late last year when I was in the UAE,

4:41 I felt like everybody knew they were number one.

4:44 And what's really interesting is they

4:45 often know exactly what their percentage is.

4:48 But as you say, you know, in December the UAE was at 64.0.

4:52 Now it is 70.1.

4:54 Yes.

4:55 That is accelerating growth in so many ways.

5:00 Yeah, certainly the UAE is doing

5:03 something that is outstanding from that perspective,

5:06 from the growth that we see.

5:07 So, the US is finally moving up the ladder.

5:11 It went up three points.

5:13 I think it's now 21st in the world instead of 24th.

5:17 It's sort of remarkable in some ways that the United

5:19 States has not yet broken into the top 20.

5:22 But you're now seeing more movement.

5:25 What do you make of that?

5:26 Yeah, it’s one of the countries that actually, in the last,

5:28 in this quarter, has grown the highest, at least number of positions.

5:34 We see a gap.

5:35 And that's one of the reasons why we are doing a deep dive on the US.

5:38 We see within the US, we see also big disparities.

5:42 You have places that are doing really well, places that are doing not as good.

5:48 And I think we have that disparity between the rural

5:51 population and the urban population that in the US is big.

5:55 I want to come to those gaps, both the north-south and the urban-rural.

6:00 But before we do, let's talk about a place that seems to be closing the gap.

6:06 One of the things highlighted by the AI for Good

6:10 Lab in this report is the growth in Asia.

6:12 Yes.

6:13 And you've really dissected that, but it's, it's fascinating.

6:16 I was in Japan and Thailand, two of the three countries in Asia,

6:21 that, together with South Korea, have been growing the most this past quarter.

6:26 And you do a deep dive on Japan

6:28 to try to understand what is driving that growth.

6:31 What is the single biggest factor?

6:33 I, we think that the biggest factor there,

6:35 similar to what we saw in South Korea, is this, is language.

6:39 In the sense that a lot of these models that were,

6:41 were not doing as well in Japanese a year ago, two years ago.

6:46 Now we are seeing that that gap,

6:48 the performance in these models in English versus in Japanese,

6:52 for example, they're almost on par.

6:54 Which means that now these models are much, like,

6:57 it's much easier for, for people to use

6:59 these models and like feel that they are useful.

7:02 So, I think that that's one of the main

7:04 reasons why people start using these models,

7:06 and they'll say they can solve problems that before they couldn't solve,

7:10 and they—they become more as a user versus before.

7:13 What is contributing to the improvement?

7:16 Well, I think it's the improvements I think

7:18 that a lot of these companies are investing in these markets,

7:21 are improving the models similar to what we saw in South Korea.

7:24 Clearly they realized there's a gap.

7:28 And I think companies like OpenAI, like Anthropic,

7:31 are improving their performance in those languages.

7:35 And that is clear from the, from the results

7:38 in some of the tests that, that we run.

7:41 I know you do other work in this space in the AI for Good Lab,

7:45 and other parts of Microsoft have been

7:47 working on linguistic capabilities really around the world.

7:51 What do you see as the biggest barriers?

7:53 Is it just a shortage of data in the local language or is it something else?

7:57 Yeah, it started with a shortage of data.

8:00 If you are in a country like the US where English is the native language,

8:03 like 50% of the content of the web is in English,

8:06 makes it relatively easy to train a very good model.

8:09 And something similar happens even with French or German.

8:12 Once you pass certain, certain languages, that's no longer true.

8:19 And we, we still have- We live in a world where you

8:22 have countries where they don't have any access to these, these language models,

8:28 because these models are not trained on those, on those languages.

8:32 That was not the case in South Korea and Japan,

8:35 but the they are not low resource languages because

8:37 they still have a good portion of the web.

8:40 But it’s not, not on par with what you see in English or German,

8:44 for example, or Spanish.

8:45 So, now there is an investment to make sure that these models

8:48 can do well in those languages and that is happening.

8:51 And we see that in the results.

8:53 And you talk in the report about data improving performance.

8:58 And then as performance improves, demand increases and you start seeing

9:03 these countries really scaling up deployment.

9:06 Can you say a little bit more about that?

9:08 Yeah.

9:09 Clearly what we see is a very good correlation between the, the the performance

9:15 in these languages from these language models and the people using these models.

9:21 I think we also saw it in the early days, like even GPT-3.5.

9:25 When GPT-3.5 was a very good model, but it wasn't,

9:28 wasn't what we see in GPT-4 or GPT-5.

9:32 Yes, you could use it for editing things,

9:36 but you couldn't solve many of the problems that you could solve today.

9:38 I think that as soon as people see the power of these models and they're doing,

9:42 they can start using it for multiple other purposes,

9:45 and that brings more adoption.

9:48 Well, I saw one of these interesting examples

9:51 of scale deployment when we were in Thailand in April.

9:54 And as you show in the report, Thailand's one of the three countries in Asia

9:59 where AI grew the most in the first quarter.

10:01 Thailand has recently applied to join the OECD.

10:06 And when you do that kind of application to join,

10:11 you have to show how you're going

10:12 to conform your domestic laws to OECD standards.

10:16 That's typically a process that takes three

10:18 to five years just to prepare the application.

10:21 If you are not a native English speaking country, and Thailand's not.

10:26 So, what they had to do was take

10:28 70,000 laws in Thai and translate them to English.

10:33 And then compare them to about 270 different OECD standards.

10:39 In this case, it didn't take them three years or five years.

10:42 It took them three months.

10:44 With a team of five people, five lawyers who work for the government.

10:49 And the difference was AI.

10:51 They used AI to translate 70,000 laws from Thai to English,

10:57 and then used AI to do the comparisons.

10:59 “Compare these laws to these standards.

11:01 Where are there gaps?

11:02 Where are there examples of what other countries have taken steps to then

11:08 close those gaps to meet the OECD's requirements?” I thought it was fascinating,

11:12 because in so many ways it not just accelerated

11:16 a process for the benefit of everyone in the country,

11:20 but it eliminated a lot of what we, I think, rightly think of as drudgery.

11:24 All of that translation, all of that laborious looking at one thing,

11:29 looking at another and comparing it.

11:32 These are things that AI is very good at.

11:35 Especially these generative models that before,

11:38 before, like AI, is not necessarily new,

11:39 but it was not doing well on dealing with text.

11:42 That is the majority of the human knowledge.

11:46 Thanks to these large language models now you can do that.

11:49 And solving something like that, before it would have been impossible without,

11:52 like, a lot of effort from humans.

11:54 Now we can do it using these models.

11:57 So, Asia's growth is good news.

12:00 The US ratcheting up three steps, that's good news for the United States.

12:07 The UAE's leadership, great news for the UAE.

12:10 But there's some big gaps around the world.

12:14 Let's talk first about the north-south divide.

12:16 What did the first quarter bring in results on that score?

12:19 When we look at the, when we look at the report,

12:21 what we see is that the Global North

12:24 continues to grow faster than the Global South.

12:26 So, we saw the Global North went up in this quarter of 2.8 points.

12:31 The Global South was less than half of that—it’s 1.3 points.

12:36 Which means that the, the gap between the North and the South continues to grow.

12:41 And once you do a deep dive,

12:42 and we actually included that in the report, it’s like,

12:44 some of the drivers is that access to the internet,

12:47 access to electricity, access to skills.

12:50 We see that not only these will likely continue to accelerate,

12:54 there is a point where some of these countries will hit a wall.

12:57 I don't think it's happening yet, but we- I don't think we are that some

13:01 of these countries are not that far from that wall.

13:04 Which means that the difference between the Global

13:06 South and the Global North will continue to increase.

13:08 And that's something that yeah, it's unfortunate.

13:11 And the report does an interesting job, I think,

13:15 of breaking it down into the different layers of technology that matters.

13:20 And, you know, as you show here, when it just starts with access to electricity,

13:26 the Global North is more than 98%, the Global South is at 88.9%.

13:32 So, you get that gap.

13:33 Then you look at access to the internet.

13:35 The Global North is at 90, the Global South 65.7.

13:39 Yeah.

13:40 So, that's a 25-point gap.

13:42 You look at access to digital skills.

13:44 The Global North is at 70.

13:46 The Global South is at 48.2.

13:49 So, there's another sort of 22-point gap.

13:54 I think it shows that closing this AI gap

13:58 actually will require that we close lots of gaps.

14:01 The electricity gap, the internet access gap, the skilling gap.

14:06 It just goes to show how much effort it's going to take.

14:10 And the bigger problem there is that closing-

14:13 Once you have the infrastructure in place, adoption is really is not difficult.

14:18 But again, this is why, explains why,

14:21 less than three years that, a bit more than three years that this started.

14:24 We see a significant portion of the world already using it.

14:27 Because once, as long as you can speak your like,

14:29 speak your language, it's relatively easy.

14:33 Having access to the internet,

14:34 having access to digital skills, having access to electricity,

14:37 the investment that these countries willing

14:39 to do is significantly higher than having like

14:41 that once- Like that part is going to be the difficult part of the adoption.

14:45 Yeah.

14:46 That gap persists.

14:47 It's even getting wider.

14:50 Your US report actually shows, interestingly enough,

14:55 a gap of similar magnitude within the United

14:59 States between urban counties and rural counties.

15:02 Urban counties are how much more than the rural counties?

15:05 Yeah, so, we see similar to what we

15:06 see within the Global North and the Global South, interestingly enough,

15:09 almost in a very similar ratio, the urban areas in the US have around half

15:16 of the AI diffusion that you see in metropolitan areas.

15:20 Rural is half of urban.

15:21 Rural is half of urban.

15:23 When we look at some, a lot of counties in the US,

15:26 they have lower AI diffusion than a lot of countries in sub-Saharan Africa.

15:30 So- And this is not because of electricity.

15:33 They actually have electricity, they have access to the internet.

15:38 There is that gap.

15:39 And this is something that I think is worth studying.

15:42 We, right now we, we can observe the data.

15:45 I don't think we know why this is happening,

15:47 but we clearly see that divide in the US between rural and urban.

15:51 We have seen that in other countries.

15:52 The rural and urban divide is not something that is unique to the US.

15:56 But clearly when you look at the map in the US, it’s clear.

15:59 The one thing that we do know, even though there's more that we need to learn,

16:06 is that we can look at a similar map

16:08 of counties in the United States and trust in AI- Yes.

16:13 -tends to correlate with usage of AI.

16:16 Trust is higher in urban areas.

16:18 It is lower in rural areas.

16:21 We've long been saying as a company, both internally and externally,

16:25 that people will only use technology that they trust.

16:29 So, is it fair to say that's one hypothesis we're

16:32 going to have to go test now and see if

16:35 we can learn some more about what it is about

16:38 trust in AI that may be part of this story?

16:41 Yeah.

16:41 We see, like you said, the correlation is clear.

16:44 Like, the rural areas in the US have much less trust, significantly less trust.

16:49 There's almost a very good correlation between

16:52 the, an inverse correlation between trust and AI adoption.

16:56 That's a pretty good hypothesis.

16:59 The thing that is noteworthy about, in part, in my view,

17:05 is that the uses of AI in rural counties are, I think, so compelling.

17:11 I mean, you just take the health care challenge.

17:15 You and I have been looking at that recently,

17:17 and there's almost 2,000 rural counties in the United States.

17:22 And yet 45% of them have five or fewer doctors.

17:27 There's 198 of them that have no doctors.

17:30 And we already see AI, including AI services from Microsoft,

17:35 from Nuance, being used by doctors to be much more productive,

17:40 to capture, the essence of a conversation between a doctor and patient,

17:45 to free the doctor up to see more patients.

17:48 The more acute the doctor shortage, I think,

17:52 the more compelling the need is to put

17:54 AI to work to help doctors see more patients.

17:58 But obviously that hasn't necessarily translated

18:01 into more trust in AI at this point.

18:05 Yeah, health care in general,

18:07 but particularly in areas where they don't have any any other solution,

18:10 I think that is a clearly a game-changer.

18:13 Similar happens to areas like agriculture, like,

18:16 using AI to help on the agriculture.

18:18 For example, reduce the reliance on fertilizers, make it more efficient.

18:23 We see a lot, many more use

18:26 cases that are extremely compelling for rural America.

18:29 But yeah, we we still don't see the the adoption of this technology.

18:32 Well, yeah, and the other one that you and I have been talking

18:35 about that the AI for Good Lab has been working on is fighting wildfires.

18:41 I mean, the ability of these AI-enhanced

18:43 cameras that you've been directly involved

18:45 in, in California to be able to detect

18:49 and identify smoke patterns that quickly show wildfires.

18:53 And, as you have been showing me, last year in the United States,

18:59 wildfires destroyed an amount of acreage equal

19:03 to the state of Massachusetts in size.

19:05 So, yeah, this is a great example of where AI can,

19:10 you know, put out fires, save homes, save lives.

19:14 It feels like we have an opportunity

19:18 for a broader conversation about how AI can be

19:21 put to work in ways that will genuinely

19:25 serve the needs of rural communities in this country.

19:28 Yeah, wildfires is a great example of the conversations that we need to have.

19:32 And I think we just started on that.

19:35 Clearly the case of California,

19:36 we would love to actually bring that to other states too.

19:39 I think that California is kind of ground zero for wildfires,

19:42 but wildfires has been affecting a significant amount of other states too.

19:46 And other countries.

19:47 Yeah.

19:48 It's a global issue.

19:49 As we continue to mature this report,

19:53 really develop it, we're not only now doing quarterly reports,

19:57 global as well as county by county in the United States,

20:01 hopefully county by county in some other countries too.

20:04 But you're now starting to get

20:06 to the point where you're also analyzing different sectors.

20:09 You started with software coding.

20:11 Why did you choose that as the first?

20:13 Well.

20:14 Software coding is clearly one of the areas

20:17 that AI is already seeing a huge improvement.

20:20 I see it in my team, where everybody in my team now is using

20:24 these tools to help them do software development.

20:27 We move- I think that what the world observed in November 2022 with ChatGPT,

20:34 we have a very similar moment that happened in December 2025,

20:38 where suddenly these models the, either

20:41 the Anthropic models or the OpenAI models,

20:43 through technology like GitHub Copilot,

20:47 it allows now the software developers to start coding in their own language,

20:51 whether that’s English, Spanish, or Mandarin.

20:54 And I'm using it myself.

20:55 And the, the improvements in productivity that we see is huge.

20:58 So, we wanted to look at that data.

21:00 It’s like, hey, clearly we are

21:02 seeing an amazing moment for software development,

21:05 is that being translated into code?

21:07 And we, through the GitHub data that is an amazing data source that we have,

21:12 we’re already observing that, right?

21:14 So, we see huge improvements in the amount of repositories.

21:17 These are projects in GitHub.

21:20 For you to be aware like, in the last six months we see more repositories

21:23 that were created in GitHub than in the first, almost first ten years of GitHub.

21:29 So, clearly we are seeing a huge

21:32 increasing productivity in the software developers’ side.

21:35 So, people are I'm sure familiar with or have heard about,

21:39 you know, Anthropic’s model Claude,

21:40 and people using it for coding, OpenAI's model.

21:44 But part of what you point to is just the evolution of GitHub.

21:48 It's no longer just a place where people store code

21:51 and make it available to a team in a repository or repo.

21:55 Even GitHub Copilot isn't just a tool

21:58 that people are using for AI to write code.

22:03 Talk a little bit about the evolution of our own

22:06 GitHub service and what that means for this.

22:08 Now, for the first time, and I think that that “Aha” moment happened

22:12 to me in last December using GitHub Copilot,

22:15 was that I no longer needed to actually write code in in Python, or C, or C#.

22:21 I was starting to use English.

22:23 And I think that, that completely changes the, the dynamics of coding.

22:30 And I think that it’s,

22:34 especially for the people that maybe their job was not coding,

22:38 I think that it's changing discipline.

22:40 Instead of writing a spec you're building

22:41 a whole prototype just by coding it in English.

22:46 This is, I think is, is going to change the dynamics of the projects

22:53 from idea to, to bringing ideas to life.

22:59 I think we're just starting to see that impact in society.

23:02 And I think that that impact is going to be huge.

23:06 You look at all of the tools that are coming together on a service like GitHub,

23:12 what does that tell you about what the future

23:14 of a software developer job is starting to look like?

23:18 I would argue we still even need more software developers.

23:21 Like, I would say majority

23:23 of the people will become software developers, correct?

23:25 So, in the sense that no matter what what your job, whether you're a lawyer,

23:29 an architect, or an accountant,

23:31 your interaction with a computer will be through coding.

23:34 You're not going to be coding in Python.

23:35 You're going to be coding in English.

23:36 But that notion will still, will still be there.

23:40 The ideas of building software, the idea of having ideas.

23:44 And I think that's going to become even better,

23:47 like more impactful, even after 30,

23:50 40 years that people have been trying to make

23:52 sure that society and that kids are coding,

23:56 around 0.5% of the population know how to code.

23:59 So, it's a very niche.

24:01 I don't believe that's going to- Sorry,

24:02 I believe that's going to change dramatically.

24:05 There's obviously a big debate with wide-ranging views across the population,

24:10 across different experts, about the impact that AI will have on jobs.

24:16 And yet right now, we're still seeing growth

24:21 in software development jobs in the United States.

24:24 In fact, the report points out that the US

24:28 Department of Commerce last year reported record growth.

24:33 8.5% increase.

24:34 2.2 million people in the United States employed as software

24:38 developers even while the growth of AI for coding was exploding.

24:44 You talk about the different economic factors that you see at work,

24:48 at least at this point in the development of AI.

24:52 You have, you know, a strong background in economics

24:55 as well as a strong background in code and AI.

25:01 How do you analyze the economics so far?

25:04 I have an economist on the team, and we’ve been discussing this.

25:06 Because, like, the first position that a lot of people are thinking,

25:09 “Well, now everybody can become a software developer,

25:13 does the productivity increase significantly?

25:15 We will need less software developers.” That's the first,

25:18 I think, perception that a lot of people have.

25:20 And that is, that is true if, if you have, a market that is fixed.

25:26 For example, that’s what happens in agriculture, right?

25:28 So, you have a kind of a fixed amount of land, you increase productivity.

25:33 Back in the early 20th Century, you increase productivity.

25:35 The amount of jobs that went into farming actually start decreasing.

25:40 But in the case of software development,

25:42 the sky's the limit, in the sense that you can grow.

25:45 And when we look at the last 30 or 40 years, this is not the first time that we

25:49 see a big increase in productivity in software developers.

25:52 Like, we used to code in assembly.

25:53 We used to code, in FORTRAN, in COBOL.

25:58 Every time that we saw an improvement in productivity,

26:03 we also saw more jobs in the area.

26:07 Because the fact that you can build more stuff

26:10 makes it more compelling for people to use these tools.

26:14 And so as as long as we think that's elastic- Right.

26:18 -this should actually translate into more jobs.

26:21 And that's what at least the data is showing.

26:23 We don't know what's going to happen in the next five years.

26:25 Right now we are seeing growth, and at least according to economists,

26:30 like this— as long as this part is elastic,

26:32 the fact that more improvements in productivity might drive more jobs.

26:36 That happened in the past, too.

26:38 And it will be interesting to see.

26:39 Obviously it's one of these classic questions.

26:41 Time will tell.

26:42 I think it's interesting when you look back in history and, you know,

26:46 in terms that almost anybody I think can appreciate,

26:50 the invention of the washing machine massively

26:53 reduced the amount of time to wash clothes.

26:57 Before the washing machine, it would take about six hours to, in effect,

27:00 clean to, in effect, clean what we now think of as a load of laundry.

27:04 And then that eventually fell to about 30 minutes,

27:07 and most of it was time where people could put the laundry

27:11 in the washing machine and walk away and do something else.

27:14 But it's the exact same point you're making.

27:16 The first thing it did was improve demand for clean clothes.

27:21 Yes.

27:22 Peoples’ whole expectation was that they would have

27:25 clothes that would be washed or cleaned more often.

27:29 Whereas there was an explosion, about a tripling, of the washing of clothes.

27:35 As software is cheaper to produce,

27:39 there is then an opportunity to use more software.

27:42 I think one of the things that will be interesting to see is

27:46 how this translates not only into the number

27:49 of people employed as software developers,

27:52 but first, as you point out, the nature of the work, you know.

27:55 And we're seeing that across our industry,

27:57 but also the types of companies that people work in.

28:01 You know, we're seeing that in some ways,

28:04 software development jobs are migrating to some degree

28:07 out of the largest companies to more companies, and not just tech companies.

28:13 And this has been going on for decades, but even more perhaps now,

28:18 where, as we have long said, every company is a software company, every company,

28:25 I think right now might be experiencing an increase

28:27 in demand and an increased ability to hire people,

28:30 because AI has reduced the barrier to entry,

28:34 who are creating the next generation of software.

28:38 One big change that I think is interesting, Brad,

28:40 is that a lot of times people think about,

28:42 software developers or even computer scientists,

28:44 as people that need to know a programming language.

28:48 And I think that that is the wrong approach.

28:50 Like, a software developer or someone that knows how to code

28:53 is someone that can actually communicate with computers to automate process.

28:58 Whether that's in English or Python doesn't matter.

29:00 So, I think it’s the wrong approach to think, it's like, “Oh,

29:04 we will need less software developers because now you can code in English.” No,

29:07 no, you will have more software developers.

29:09 It’s going to become easier to code.

29:11 It's already easier to code.

29:12 I think it's a really interesting point that you make,

29:16 because it's not just people who are working full time to develop software.

29:20 Maybe a great many of us,

29:23 many white collar workers of all kinds of backgrounds and professions,

29:29 will be, as you say, communicating with a computer and in effect,

29:34 creating something that is manifested in code,

29:39 that we put to work, especially in agentic AI, to help us do our jobs.

29:45 And we're already seeing this.

29:47 I mean, I'm blown away when I see some of our employees,

29:50 some of them are young, some of them are of all ages,

29:54 and they are doing things in 2026 that I don't think many of them

30:00 could have possibly imagined they'd have the ability

30:03 to do two or three years ago.

30:05 And yet they're doing it in ways that change legal work,

30:10 public policy research, communications, you name it.

30:15 There are people at work today now starting to use

30:18 AI to change the way they're doing their job.

30:20 And I think that is the real manifestation, in part,

30:23 of how we can put AI to work in white collar

30:28 professions to make ourselves better at whatever we want to do.

30:33 In a way, if we if you think about what happened in the 80s and 90s,

30:36 with Windows, and even with tools like Excel,

30:40 back in the 80s, for example, not the majority of people, like,

30:42 in order to use a PC they need to understand a disk operating system.

30:46 It was kind of like almost like coding in many ways.

30:49 Windows democratized that, because it made it much

30:52 easier for people to interact with a PC.

30:55 You didn't need less people,

30:56 suddenly everybody became like, started using these tools,

30:59 started using an operating system in a way that it was easier.

31:03 The same now is, I think the same is happening with coding.

31:06 Where before, it was something that was very niche,

31:09 now it’s going to become that new Excel,

31:11 that new tool that will allow them to increase

31:14 the productivity of things that they couldn't do before.

31:16 Well, that's a lot of interesting stories for the first quarter of 2026.

31:23 It's going to be interesting to see what the second quarter brings.

31:26 Which of these trends continue?

31:28 Which of these trends change?

31:30 Look forward to comparing notes three months from now.

31:33 Juan Lavista, thank you.

31:34 Thank you, Brad.