Women WithAI: Surfing the AI Wave: Nisha Talagala on Empowering the Next Generation Through AI Education

Nisha Talagala: 00:00:02

So I can tend to describe AI as riding the wave. The AI is like a wave, right? It's a very, very powerful, large wave. You can either be on top of it or drown in it. There really isn't that much left.

But if you can surf the wave, you can do a lot of cool stuff.

Joanna Shilton: 00:00:15

Hello and welcome to Women WithAI, the podcast dedicated to amplifying the voices and perspectives of women in the field of artificial intelligence.

AI Trends to Expect in: 2025

So I just had to invite her onto the show. I'm really looking forward to speaking to her today. But before we jump into the podcast, let me tell you a little bit about her.

Nisha Talagala is a recognized leader in the operational machine learning space.

She has over 20 years of experience in enterprise software development, technical strategy, product leadership, leading to her work with AI, machine learning and distributed systems. She's experienced in introducing technologies like artificial intelligence to new learners, from students to professionals.

She's also a committee member overseeing an AI education initiative across 100 schools in Sri Lanka.

Neesha co founded Parallel M, which pioneered the MLOps practice of managing machine learning and production for enterprises and the company then went on to be acquired by DataRobot. She's also led groundbreaking projects at companies like Intel, SanDisk and Fusion IO.

Nisha holds a a PhD from UC Berkeley where she did research on clusters and distributed systems, holds 75 patents in distributed systems and software, has over 25 peer reviewed research publications, and is a frequent speaker at industry and academic events as well as being that contributing writer to Forbes and many other publications.

Above all else though, Nisha is passionate about making AI accessible to everyone, helping them navigate the opportunities and challenges of this transformative technology. Nisha Talagala, welcome to Women with AI.

Nisha Talagala: 00:02:01

Thank you very much. Very happy to be here.

Joanna Shilton: 00:02:04

Oh, it's great to have you. Thank you for coming.

And I, I do want to ask you to tell us about what inspired you about your passion for technology and your journey into AI, which ultimately led you to found AI Club. But first of all, what are clusters and distributed systems? Will that become clear as we speak? Because I, I don't know.

Nisha Talagala: 00:02:24

Yeah, absolutely.

So, so what happens is that, you know, originally when computers were created, the way you got more done was that you built a bigger and bigger computer Right. And these were called supercomputers. And some of these are honestly bigger than our fridges, right. And they had their own like liquid cooling.

So if you walked into one, you know, looked at one of them, you could actually see the liquid sloshing through, trying to keep it cool.

And then what happened is that people, particularly people at Berkeley, realized that maybe we don't need to build bigger and bigger computers, we simply need to get a bunch of computer to work together. So that's what a cluster is. A cluster is basically a bunch of computers that are going to try to be like a big computer by working together.

And so, you know, and back then, by the way, that was a pretty revolutionary idea. Like at Berkeley we had a cluster of 100 computers. And this was like unheard of, you know.

And by the way, it's not that Easy to get 100 computers to work together. Like so the way I explain this to kids is imagine if you had 100 people and you are trying to get them to do something.

First of all you need a megaphone because otherwise half of the room will not hear you. The second is at any point in time, some people are ahead, some people are behind, some are asleep, some are sick.

I mean, this is the way it is, right? And the more people you have in the room, the more of that that's going to happen. Same with computers. It's hard to keep them organized.

Somebody's dying, someone's overworked, someone is under worked. And so a lot of the things about clusters is about that. And it was a hard enough problem at 100 computers.

But these days if you look at Google, Google has millions of computer and this is how it works. So Google doesn't do everything. It does because it has one enormous computer that's several buildings big.

It has little computers, rows and rows and rows and rows of them. So that technology was actually pioneered by projects like Berkeley.

And Berkeley in fact led to one of the first search engines in the world, a company that got acquired by Yahoo, that is no longer nobody knows about, but we were literally one of the big first search engines. So anyway, so that's what very strongly that is what distributed computing is.

And we do it every day these days, anytime you log into Google, you're using a cluster distributed computer. You just don't know it.

Speaker C: 00:04:29

Wow, that.

Joanna Shilton: 00:04:30

Yeah, that blows my mind. I hadn't ever really thought about it. And now, yeah, I'm going to be thinking about it all the time. Thank you. Well, brilliant. So now. Yeah.

How did you get interested in all of this because you're obviously passionate about technology and. Yeah. And using. Well, AI. I'll let you tell it in your own words.

Nisha Talagala: 00:04:46

Sure.

So to be honest, I mean, when I was little, like, you know, I don't know, I want to say about 7 or 8 years old, maybe a little older, there was a TV show that I used to watch. It was called Small Wonder. And the idea was that there was a.

This was a family and there was a little girl and the little girl was a robot and only the family knew. And if you go look it up, it's a very old TV show, but you can see it. And so.

And I was like, you know, I want to build myself one of those, you know, And I had all these dreams that I would have one and no one but me would know that it was a robot. And I would, you know, keep her in the closet or like fix her, and then she would come out and play with me and stuff like that.

So that's actually how my first interest in AI came about. And then when I went to study computer science for a long time, I mean, I did some AI when I was about 17, 18, but I didn't do AI for a long time.

That's how I got into the Clusters. I had an amazing advisor at Berkeley. I kind of followed his directions and did some really cool work. And that I'm very happy about in Clusters.

uite a while. But then around: 2015

And I'm super happy that I did that. So I've been in AI now again for about 10 years. And it was certainly a wonderful time because it was before all the latest craziness.

But the latest craziness is just so much fun. So anyway, that's a little bit about my. So these days, I mean, I haven't built a robot yet, much less one that can mimic a human.

But I found that there are so many other cool things you can do with AI that don't have to do with fooling humans, thinking that your best friend is real when she's actually a robot.

Joanna Shilton: 00:06:17

I like that. Not yet. You haven't built it yet? Not yet. Well, AI Club is such an exciting initiative. I mean, for anyone that isn't aware of it.

Can you just explain it in your own words, what you do?

Nisha Talagala: 00:06:28

Absolutely.

And I think maybe a little bit of this also help to Understand the history, because it's not just my history, it's the history of all the people who work at AI Club. The leaders of AI Club is we all started as AI professionals.

Like my previous company, ParallelM, we were helping big banks, hospitals, insurance companies build AI situations where a lot of trouble happened if they went wrong. Legal trouble, human trouble, you know, lots of trouble. And so we already knew how to do that. We knew how to build big AI, safe AI systems.

But then we all kind of realized that honestly, this is a technology that everybody needs to learn. And it I think helped a little bit that we were parents.

So I, at the time, you know, I started teaching my daughter AI when she was nine and I realized a couple of things. One, she could learn and B, if I did not really put an effort in, it got boring and tedious really fast.

And you know, nine year old has no patience for boring and tedious, right. I became her data intern. She was frustrated, I was frustrated. It was a mess, right?

But I kind of realized, hey, this is possible, but you do actually have to think about how to do it well and to engage them. And so that's sort of. And so when we started teaching AI to kids, a lot of people looked at us funny and so what do you mean?

And I even had one dad say, I don't understand this stuff. How do you expect my child to understand it? And then he comes to me a couple of months later saying, how does she understand it better than I do?

So, so, so I think the key was first realizing that this is actually a good idea, you know, and now, by the way, you know, Asia, for example, starts in third grade. They start teaching kids AI in third grade. So. And we were like, heck, we were more conservative. We started in middle school.

Joanna Shilton: 00:08:04

I think people were scared of it, weren't they, to begin with? Or didn't want or worried what it might do. So didn't, didn't know how to explain it.

That's the other thing we, you're right, grown ups don't understand it. People that aren't using it, how to.

Nisha Talagala: 00:08:16

Explain it and how to help kids. And pretty much anybody. It's not just kids help people.

So one of the things about AI is that it is sort of both extremely deep and in everyone's lives at the same time.

Now if you go back to like the distributed systems and cluster question you asked, you know, most adults could spend their entire life benefiting from clusters without ever needing to know what they were. It's not like, you know, if you don't want to know what a cluster is. You are fine.

You know, it's really never, ever going to become a problem for you. And most computer science is like that, you know, unless you care about it, you can safely ignore it. But AI is not like that.

AI is in your face every single day, you know, and it's, you know, it's in the way your doctor treats you, it's in the way your bank treats you, it's in the way that you go onto the Internet. And every question is about more as much about you as about the question.

Speaker C: 00:09:08

Yeah.

Nisha Talagala: 00:09:09

So, for example, if I decide that I want to get an Uber or some kind of a ride share, the price I get isn't just about the time and the trip. It's also about me. AI has learned what I'm willing to tolerate.

And if I'm the kind of person who will pay more for my convenience, it will charge me more so because it has learned. So this world, like, where the pri. Everything that you deal with is now not just about your question, it's also about you.

And these systems know more about you than you realize. So when you have those kinds of situations, it's important that everybody at some way understands it.

And some people need to understand different things. Some people understand more or less. So we kind of realize that.

And I think one of the things that I feel we've done really, really well is we've adapted how people learn it to what is right for them. Not everybody wants to learn probability and statistics, and neither do they need to. Some do, some don't.

Some people need to understand how it interacts with their daily lives. Some people are like, you're a doctor. You need to understand an entirely different set of things about privacy and health.

And are these things actually right? If you're a writer, you want to understand other things. So we have.

I think we've put a lot of effort into making it the right fit to everyone because it has to be the right fit for everyone. So that's kind of like what we do at AI Club is we call ourselves 8 to 80, practically speaking. We started with my daughter.

She was maybe 11 when we started. So we started with her and kids like her. Then we went from that to high schoolers. Then we went from high schoolers to elementary. Then we went.

We've always worked with professionals because we are professionals. So we've always worked with professionals. And most recently, we are working with college students and students just graduating from college. So.

So you can see kind of like. And Every approach is a little different, but it's all about AI.

Joanna Shilton: 00:11:02

Yeah, because you're right, it's everywhere. Like, even today, I was, I was emailing.

I was emailing you this morning and it suddenly, my Google Mail had changed on my phone and it was like, oh, can I help you? And it popped up with something and then it did on my. I was sending a message to my mum on, on my iPhone again.

I was using Messenger, Facebook messenger. And it's. Suddenly there's a box and I was like, what? This wasn't there yesterday. Like, it's moving so quickly and I'm, I'm.

I'm more used to it than, say my parents are or, or other people that, that kind of don't. Aren't taking any notice of it. But you can't hide from it anymore, you can't ignore it. So you're, you're right. It's.

It's so important that children are being taught about it. I mean, they can then teach their.

Nisha Talagala: 00:11:45

Parents and then there's that. Yes, absolutely. So, so, yeah, so that's, I think, you know, that's kind of how we got into what we did.

And I think one of the reasons why we're so good at it is because we actually came from the professional world. Right. So we know what is actually out there.

Speaker C: 00:12:01

Yeah.

Nisha Talagala: 00:12:01

We also understand the need for everyone to know it. And we. The effort we have put in as to how to connect those two, two things together. Right.

Is how do you take what is true, what is state of the art, what is absolutely bleeding edge, but not scare people with it, but help them understand the things about it that are important for them to thrive?

Speaker C: 00:12:19

Yeah.

Joanna Shilton: 00:12:21

And why do you think it's important? I mean, is it so people use it without being scared or, you know, they wouldn't use it if they don't understand it?

Or is it so that everyone's kind of learning at the same time?

Nisha Talagala: 00:12:35

So I think it depends a lot on the audience, but there are a few elements. Right. So I'll talk about the kids, for example. So one thing we have learned from the kids is that the kids have really infinite imagination. Right.

You know how women life teaches us to not imagine at the end of the day. Right. You know, we get beaten down and we're like, no, I'm pretty sure we're not supposed to do that. Our first question is, are we allowed to do it?

Second question is, do I want to do it? Can I do it? We go through all these before we actually do it. Kids are not like that. Kids are like, oh, I'm going to do that.

I didn't even realize it was not possible. Right.

And so that I think is most important about teaching kids AI is that it gets them into this mindset that maybe there's a bigger problem they can solve if they have the right tools. And that is the key.

So you put the tool in their hands, you give them some examples of how the tool works, you have them build a few things that you know about, and then you ask them, what do you think you should do? So one of the things that we do in every single class, so we do two things in every single class we teach.

The first is the students build an AI on their first day. They build one literally from scratch. This is not about using some tool that is out there.

They build one and they tweak it and they understand what it's good at, what it's not good at, how they can make it better. That goes a long way towards sort of setting their relationship. The fear goes away, the sense of empowerment comes in. Right.

They realize that they have agency in this situation and things like that. And that's where we begin. Then every class we teach, every course, it's a multi class course.

Every course we teach, every student has a custom project. Now they get a lot of guidance, of course, tons of ideas and massive library of projects from the past.

They can pick one if they want to, but they can create their own. Then what happens is that by the duration of the class, they've gone from not being scared to understanding to creating.

That creating is a very big part because now it's like, okay, I can actually do something with this stuff. Right. And one of the beauties of AI in the fact that it's literally everywhere is that all kids are different. So find your joy if your joy is in.

Like, for example, we've had, we had a, you know, a young lady who, she's a ballet dancer. She built a tool where you can, you know, take a video of yourself doing ballet and it will tell you what is wrong with your pose.

Speaker C: 00:14:52

Wow.

Nisha Talagala: 00:14:54

And. And you know, if your pose is wrong, you could injure yourself. It's not. I mean, there are many, many things. So it will.

Your movement is a little slow here, it should be smoother. Your position of your leg is wrong. You could hurt your back. You know, stuff like that.

Speaker C: 00:15:05

Yeah.

Nisha Talagala: 00:15:06

Then similar things have been done for violin, for tennis serves. At the same time, we have studied students who are looking at cancer genetic markers. You know, I think so. You Know, and stuff like that.

And so the key there is that you. We are not. We will teach you the AI, but we don't mandate what you do with it as long as it's ethical.

We let you merge your passion with it and help you sort of like. So I can tend to describe AI as riding the wave. Like, the AI is like a wave, right. It's a very, very powerful, large wave.

You can either be on top of or drown in it.

Speaker C: 00:15:38

Yeah.

Nisha Talagala: 00:15:39

There really isn't that much left. But if you can surf the wave, you can do a lot of cool stuff. Yeah, you teach them to surf the wave.

Joanna Shilton: 00:15:47

I think that's such a good analogy because it's almost like the people that are burying their head in the sand or, well, they're just. They're not looking at the wave. They're not looking and it's just going to crash over them and just knock them over because they're not expecting it.

That's because I think there's so much, you know, about, oh, it's. It's terrifying. It's going to take your jobs, it's going to do this, that and the other. And it's. You're right.

Like, this is coming at it from a completely different angle and starting, you know, with educating people about it.

It just makes so much sense because then, you know, how to spot, I suppose, the scary bits and the bits that people don't trust because, I mean, it's really interesting because you're saying it's use it. You know, the kids are sort of using it and they're using their imagination. Start with.

Because a lot of the time people are like, oh, well, it's going to take away imagination. You know, if art and music is just going to be written by AIs.

But I guess that's where you've got to have the humans that have written the AIs that are then nurturing them and using them, as you say, as a tool. You're not. We're not just going to sit back, I hope, as a race, and just be like, oh, AI, over to you.

Nisha Talagala: 00:16:47

Yeah. So, like, for example, and this is very important because, I mean, we.

I mean, we make, you know, we teach AI ethics in every room, but sometimes we don't teach it as a topic. We just make sure they appreciate the ethics through experience. So I'll give you a simple example. Like, for a.

Like for a young child, the first thing we do is we show them how to build an AI. It's a very simple AI. It detects Whether you're happy or sad by the stuff you type. Okay? And so.

And it was, you know, sourced by, you know, kids their age. So, you know, it'll think that chocolate is happy and broccoli is sad and, you know, this and that.

And so it's very, very kid, you know, natural to contain anything that, you know, is inappropriate for kids. Then what we do is we let them loose and ask them to teach it whatever they want. So, you know, someone will teach that rabbits are good.

Someone else will teach that their little brother is the source of all evil. You know, I mean, they will teach anything, and they'll have a glory of a time with it.

And they will also see that the things they teach it change its way of thinking, you know, and what it takes to get the AI to believe whatever you want. And then I always ask them a question. Did the AI ever ask you whether what you taught it was true? And the answer is no.

The AI will never ask you whether what you taught it was true, which means you can teach it anything. And then I ask them, okay, who's responsible for making sure the thing is behaving? And the answer to that is you. Right.

So what is your role in all of this? It's my job to make sure the thing behaves. One of the many ways I do that is by not teaching it bad things. And so it's very important because it's.

So did we teach them ethics? Yes, but we didn't teach it in the abstract.

By simply realizing how easy it was to teach the AI any bad thing, they immediately understand their role in this situation. It empowers them, and it also shows them the right way.

And now they're in a position where they can talk about bias, about fake news and this and that, because they can appreciate the process that leads to it and the trouble that it creates.

Joanna Shilton: 00:18:47

And they'll get better at spotting it than, say, their parents or people that aren't. I want to come to your school now. I want to join the AI Club.

Nisha Talagala: 00:18:58

You see how it's kind of like it weaves in?

Joanna Shilton: 00:19:00

Yeah. And so because you're doing. If you're doing it at the schools in Sri Lanka, have you seen a difference?

Like, is sort of doing it to a different audience? You know, is it different there to. It is in the US or.

Nisha Talagala: 00:19:13

I believe it is in many ways. But I can tell you that one of the easiest ways to talk about the difference is that we focus on problems that matter to them.

For example, the kids that are building custom projects In Sri Lanka, some of them are building them on, detecting pests that are unique to that part of the world. There was one kid who did predicting used bike prices.

It turns out the used bike industry is a very big deal in Sri Lanka, and there's a lot of people trying to figure out the right used bike price. So you can see that there's a lot of stuff that you can do that's very, you know.

Speaker C: 00:19:45

Yeah.

Nisha Talagala: 00:19:46

Very unique to that environment. And there may be things, but there are things that cross over. Everybody loves Harry Potter, so like that.

So some things cross over, but, you know, you do see kind of a shift in.

And that's one of the things I love about the custom projects, is that the custom projects allow these children to do something that matters in their world.

Speaker C: 00:20:03

Yes. Wow.

Joanna Shilton: 00:20:06

And I guess doing that. Yeah, it's just putting them in good stead for the future and so. And helping to understand because there's lots of.

There's governments using it, like over here in the uk you know, the government has just announced that they want to, like, you know, instill AI in everything. I mean, how is there anything that worries you about all of that?

I mean, how, you know, in all your experiences, do you think we should be paying closer attention to anything before it happens?

Nisha Talagala: 00:20:34

I mean, so if you look at any example of a massive technology wave, it has enormous good and enormous bad. AI is probably larger than any of them.

So I'll take a simple other wave that, you know, and I can take two other waves and I can show you why AI is larger, both positively and negatively than both of them. So if you think about nuclear power, right. Nuclear power powers a lot of homes on the planet. We may not even realize it.

You know, it's a very powerful technology. But we've had our share of nuclear accidents. You know, they've been devastating.

And even to this day, keeping the planet from being destroyed by nuclear power is an active effort. There are issues, Right. There are rogue nations or nations, right?

Speaker C: 00:21:15

Yeah.

Nisha Talagala: 00:21:15

We try to control are they building centrifuge this pluton, you know, enriching that. It's not like a thing. It's not a solved problem. It's a problem that's actively being maintained. Yeah.

On the other hand, the average person cannot build a nuclear reactor. And you don't build one without leaving a ton of evidence behind.

And so that's an example of a technology that can do tremendous damage in one bad event. But it's really hard to build. Now, if I take the opposite of a car Right. A car is literally everywhere. We all have them.

Our ability to cause trouble with cars is nearly infinite, and we have laws to prevent them. But if you decide to cause trouble with a car, there's only so much trouble you can do with a car.

You can set your car on fire, you can hit someone, but you can't like eradicate a city with your car. So the problem with AI is that it has the ease of the second and the damage of the first.

Like, I can sit on my computer and create trouble for a lot of people, and I can do it without anyone being able to track me. So it has like, almost like the nuclear level damage and the car ease. Yeah, that makes sense, right?

Joanna Shilton: 00:22:21

Yeah, that's the scary bit.

Nisha Talagala: 00:22:23

And that's the scary bit. And so, but just like it has to be managed. It has a potential for a lot of good, it has a potential for a lot of bad, and it's not the AI.

One of the things that I think we try very much to instill on our students is AI is not good and AI is not bad. AI is just a computer. Humans make it good or bad. Humans are always the cause of most of our problems, to be honest. You know.

Joanna Shilton: 00:22:44

Yeah.

Nisha Talagala: 00:22:45

And so it really comes down to. And what we will do with it as a society will determine how it goes.

Speaker C: 00:22:51

Yeah.

Joanna Shilton: 00:22:52

And that's where the education comes in.

Nisha Talagala: 00:22:55

That's right.

Joanna Shilton: 00:22:55

And it's so important. Can you've just done. Because you've just started the AI club, the research program. Is that right? Is that a new. Is that a new ARM or.

Nisha Talagala: 00:23:03

No, it's. We've been doing it for a few years now. And it's definitely something we're very proud of. And it's also another.

Like everything else was basically something that a lot of people thought was really just nuts when we started it.

But one of the things that has happened that we realized this by having kids build their own projects, is there's a tremendous amount of public data out there and a tremendous amount of easy to access compute.

Like, I can sit at my computer, I can get myself a free Google account, and I can do things on their massive forms of computers that you could never imagine, at least when I was the age of these kids, that allows them to solve problems that were previously the domain of graduate students. And they have the imagination, the energy, the, you know, the aptitude, as long as it's taught them the right way.

So the research program is almost exclusively focused on high school students, sometimes middle, sometimes college. But we have asked them to really take the research to the state of the art. We teach them how to think, how to write. Right.

How to solve a hard problem.

So we have kids in the research program who are doing amazingly innovative things with climate science, with, you know, genetics, with bio, you know, bioinformatics, and also kids who are doing amazing, innovative things in sports, in music. Right.

But the idea is we want them to start early, understanding how to tap into their own brains and into the technology to really solve hard problems. That positions them extremely well as they go into college to further that journey.

Because if you think about it right now, one of the things that I, you know, I find very. It's something everyone should think about. The latest AIs. You know how when AIs are announced these days, they come with their test scores?

Speaker C: 00:24:46

Yeah.

Nisha Talagala: 00:24:48

The AI from, you know, AIs last year be it did better on the US medical licensing exam than 80% of the country's doctors. You know, they are, they excel at the scat stat, at the bar exam, every possible, you know, measure that we use.

And so now I think about it, and I think about, like, kids like my daughter. I mean, why should. What is it if, if these AIs today.

AIs today are doing better than most young people in an exam she's not even going to take for four years.

Speaker C: 00:25:20

Yeah.

Nisha Talagala: 00:25:21

Why is she even taking that exam? What does it even mean?

Joanna Shilton: 00:25:23

Yeah.

Nisha Talagala: 00:25:24

Right. So where is she going to find her job? And so the research is sort of our way to show you you can do more. Will show you how.

Joanna Shilton: 00:25:37

Yeah. Because if the AI has learned all that and done all that and you can just go, what is the answer to this? Or how do we do that?

Then hopefully that frees you up your mind to look into something else or to actually go and speak to that person about this or look into something else.

But then what do you think then about the kind of the fact that people talk about AI hallucinating or that it's learned from all the data that's out there? So now it's just going to start learning from itself. And it's not always true.

It doesn't always give you the right answer because, you know, you can say just an example in, in the office where I was working yesterday, someone said, oh, well, I asked, you know, chatgpt to refine my Twitter message. So X message. So it was 280 characters. And they said. And it said, yes, it is. And then they're like, no, it's like 350 and it's. And so What?

How do we make sure that the AI isn't just going to start fabricating stuff and making stuff up? Because if no one's learned the bar exam or done the medical, you know, seven years, then what happened? Like what? Yeah. What do you think about that?

Nisha Talagala: 00:26:38

No, no, this is. This. This, I think, is the crux of the thing.

So AIs are very, very helpful to people who can read what they create, separate the good from the bad, and build on now. But, but. So you should never trust an AI blindly, ever.

So it's really not that you don't learn these things, it's more that you learn them, but you go beyond them, right?

Joanna Shilton: 00:27:02

Yeah.

Nisha Talagala: 00:27:02

You know, and there are things that you may not learn, but it's not that you don't learn the topic at all, but maybe you don't necessarily learn. Like, for example, like, my mental math these days is awful. You know, but that's okay because I, you know, I can use a calculator.

I have calculators available in so many different forms, including on my computer if I had to. I can do mental math. I just don't. Can't do it as fast as I used to. Right. So it's really about that. It's not about black and white. I don't do this.

I don't do this. Like, I focus my energy on other things.

Speaker C: 00:27:32

Yeah.

Nisha Talagala: 00:27:33

So.

And what makes things really hard, particularly for education systems, is that gray we can do binary, we can say, you don't need to take that course anymore, you have to take this course. But it's not that simple. You have to take that course. Just maybe not as effective, intensely as you used to. And that's so hard to traverse.

Speaker C: 00:27:49

Yeah.

Nisha Talagala: 00:27:49

So a real and a simple example, I wrote a Forbes article about this that, you know, that I'm actually happy that I wrote very much, is that it's also about code. Like, AIs can write code, and their code has gotten a lot better. It's actually not bad at all, the code that it generates.

But now the question becomes, I know that it can do better because I can read its code, and I do every single time. I never run anything that I don't read. And at the same time, how did I learn how to read it? I wrote a lot before the day of AI. Right.

Now, if you look at kids, if you don't make them write it, how can they learn to read?

So you still have to make them write it, but you also have to help them to appreciate that writing may not by itself be A skill that they can sustain. So it's kind of like you have to learn to write because otherwise you won't know how to read.

And if you don't know how to read, you can't build on it. But you also have to understand that if you learn how to write, you can say, hey, I'm done. Because that skill will not get you employed anymore.

So you see how it's tricky, right? And hallucinations tie into this. I mean, AI is. Hallucinations are honestly.

I think if you talk to different AI people, they'll tell you it'll go away. And it honestly is getting better. But you also have to understand that they are also part and parcel of how the thing works.

Humans come up with stuff all the time. Sometimes you don't even realize we're making things up.

Joanna Shilton: 00:29:07

Well, exactly.

Nisha Talagala: 00:29:08

So it's one of those things that you just have to kind of. You cannot have a blind faith in the thing. So you have to find some middle ground. You cannot ignore it. You cannot use it blindly.

You have to use it intelligently. And that is what is so hard. Because it's not an easy thing. It's not an easy thing. I can do it. I can teach people how to do it.

But imagine a legion of teachers trying to teach people how to do it. Right? Yeah, it's not easy at all. And that was actually a funny thing that I saw. So there's.

So I'm not going to drop any brands here because I don't think that would be fair. But there are several companies that are coming out with, quote, unquote, AI engineers.

This is a piece of software that I can buy that effectively is meant to work like a junior engineer. Okay. And it's, it's. And so if you think about. That's very, very direct, right?

Joanna Shilton: 00:29:56

Yeah.

Nisha Talagala: 00:29:57

And I can tell you that it is 110 more than 1/10 cheaper than a junior engineer. But here's the interesting thing. Somebody posted and they hired one of these things, right? Or I don't know if hired is even the right word.

They bought hired, you know, onboarded, downloaded, whatever it is, one of these things. And they said, you know, it gave. I gave it a small task and it nearly deleted all of my company's data. And, you know, and people responded.

Every junior engineer does that.

Joanna Shilton: 00:30:20

Oh, yeah, this is why we have guardrails.

Nisha Talagala: 00:30:23

It's a right of passage for any young engineer to almost delete the company's data. Otherwise, how do they learn it? So there was like, the massive joke is like, oh, it's behaving exactly like a junior engineer.

Speaker C: 00:30:32

Yeah.

Nisha Talagala: 00:30:33

So. So, you know, so to end that really interesting middle ground.

Joanna Shilton: 00:30:37

Yeah.

You have made me feel a lot, a lot happier or kind of a lot, you know, give me lots to think about, but just that it's all about the education and people. Yeah. We need to embrace it. And you're right. We can't, you know, take a hit in the sand. We've got to be prepared to ride their wave. Yeah.

Nisha Talagala: 00:30:55

There's nothing else to be done. And the other thing to also remember is that any AI we teach right now is not the AI these kids are going to use in five years.

So we're not asking them to be experts in that. We're asking them when something new comes, they understand the fundamentals well enough to appreciate what the new thing is and surf the way.

Speaker C: 00:31:09

Yeah.

Nisha Talagala: 00:31:10

And if you teach them how to do that, then whatever shows up, they will adapt.

Speaker C: 00:31:13

Yeah.

Nisha Talagala: 00:31:14

And that's what we did. Like, the things I used these days are not the things I learned in college. Heck, the things I learned in college barely exist anymore.

Speaker C: 00:31:21

Yeah.

Joanna Shilton: 00:31:21

No, you're right. It does all change.

Nisha Talagala: 00:31:23

It all changes. You have to get used to that.

You have to accept that, and you have to learn to adapt to it, because there's nothing that is probably the single valuable, most valuable skills. You have to learn to think with tools.

Speaker C: 00:31:34

Yeah. Yeah.

Joanna Shilton: 00:31:36

So what excites you most about the future with AI? What do you. What do you see coming?

Nisha Talagala: 00:31:42

And honestly, the research program really excites me because every day I see things that, you know, I. I see young people doing things that I would not have thought were possible a decade ago.

And I see the amount of just available knowledge and power at their fingertips. Right. 99% of the kids we work with are working with open data. Very, very few of them have proprietary data. And they're able to do amazing things.

And also, the amount of open data that's out there is unbelievable.

Speaker C: 00:32:12

Yeah.

Joanna Shilton: 00:32:13

I think lots of people don't realize, do they, that once they've used an AI particular, it's out there.

Nisha Talagala: 00:32:20

Exactly. And the amount of software that's out there to process that open data is unbelievable.

So I think that really, really gives me hope because I see and through the research program, just young people truly doing incredible. Many of our research program students work on more than one project with us. So I see their thinking evolve.

Their thinking, their capabilities, how do I say? Their drive. It's interesting because we had one young man, a very exceptional young man, who recently got into mit.

And so of course getting into MIT was impressive, but that's not what impressed me about him. First year at MIT, he convinced MIT's top AI research lab to accept him. He's an undergrad.

It didn't even occur to him that he needed to be a grad student. He got them to accept him. And that's my point, is that you want to put them in a position like, oh no, no, I can do this.

Joanna Shilton: 00:33:07

You know, yeah, we all need to be more like that.

Nisha Talagala: 00:33:11

So hopefully we can just, I mean we don't all need to be that, although, you know, but we need to be more like that. To your point is that you want to feel like this is going to empower you.

Speaker C: 00:33:21

Yeah.

Nisha Talagala: 00:33:22

And that's, I think that's what I'm excited about.

Joanna Shilton: 00:33:25

And is there anything that we should be paying closer attention to, do you think? Is there anything that you not warn students but you just say make sure you think about that.

Nisha Talagala: 00:33:33

I mean, I think generally speaking I would say that anytime you work with an AI, think about who is paying for that thing and what is it that they want. Because the AI by itself is not good or bad, but it is probably extremely good at doing whatever it is that its creator wanted you to do.

A lot of AIs want to learn more about you. Whether you choose to let them is entirely your decision. But you should understand that that's what they want.

In fact, one of the things that, you know, one of the little exercises that I typically run with, you know, school age children is we take a big, I'm not going to in a name, but we will take a big website that sells everything, right? And we, I ask them, what is it that this site knows about you?

And we have an open, empty whiteboard and we fill it in and by the time it's done, we fill. So first thing is, does it know where you live? Yes, otherwise where would the packages go?

Second is, you know, do you think it knows how many people are in your family? Well, if you are buying clothes from it, it probably figured out that there's, you know, who, how you are, roughly how old it.

Does it know when your birthday is? If you or bought birthday cakes probably or any other kind of banner the. And you know, does it know where your grandma lives?

You ever order a package for grandma, a gift that you delivered? Does it know if you've moved? Yes. Your addresses has changed?

You know, does it know whether you've lost weight pretty down likely if you are, you know, if you've Been buying clothes.

Speaker C: 00:34:55

Yeah.

Nisha Talagala: 00:34:55

Right. And this and that. And this, by the way, is.

And we can fill the whiteboard with things this company knows about you without ever touching things that are legally private.

Speaker C: 00:35:05

Yeah.

Nisha Talagala: 00:35:06

This is assuming its device in your house is not listening to you.

Speaker C: 00:35:10

Yeah.

Nisha Talagala: 00:35:10

This is assuming it doesn't know your prescriptions. Right. So now you think about, what is it that it knows? Right. And then now they want, do you. Are you okay with that?

And so one of the questions I ask kids is a simple thing, which is if a company, if your grocery store tells you that if you buy a loyal, get a loyalty card, you'll get 10% off on eggs. But they will know that you buy eggs.

Speaker C: 00:35:31

Yeah.

Nisha Talagala: 00:35:32

And most of the kids are like, big deal. It's eggs. You know, there will be at least one or two children. They have no business knowing whether I like eggs. Right. And that's a personal choice.

Then we ask the next question, which is, okay, no problem with eggs. What if you buy back pain medicine off the counter? Now, some of the kids are like, I'm not sure I like that. You know, Next question.

Okay, it knows you bought back pain medicine for your dad. I probably figured out that your dad has back pain. What if it goes and tells your dad's boss? And now all the kids are like, no, no, no, no, no.

So half of it is just understanding where your comfort lies, because these systems are tune from the ground up to learn about you and find a way to use that knowledge to make money. And that could include trying to sell you things. It could include teaching about you to other people. It could include so many things.

So you be aware and you do what's right for you. So that's what I would say is everybody should just. Just be aware.

You know, it's your decision whether you are willing to let them know, you know, whether that you like eggs. That's entirely your decision. There's no, like, moral, absolute right answer to it.

But I hope that it's a decision you make yourself and it's not made for you.

Speaker C: 00:36:42

Yeah.

Joanna Shilton: 00:36:43

And that's why, if you understand it, you can put things in place so that you're not sharing that or you are sharing that, or you don't mind sharing that, but you don't want to share that. And I suppose, yeah, that's where people get scared because they don't know what to do. Nisha, I could talk to you all day. This is fantastic.

But everyone that's listening, that wants to know more about this and want to start their own journey into like maybe building AI or literacy and the resources and tools like what, what advice do you have? Where do you recommend people start? Or where can they go?

Nisha Talagala: 00:37:13

We, we. So my colleague and I did write a textbook. It's a textbook for middle school and high school kids.

But it's honestly, it's a really easy read for even adults. You might want to try that, I think. You know, Shashi sent you the link. Yeah. Go to our website. There's a lot of cool resources there.

A lot of blogs that we write. Yeah, yeah.

Joanna Shilton: 00:37:31

Or is that the fundamentals of Artificial intelligence?

Nisha Talagala: 00:37:34

You can take a look at that. And you know, and I would say maybe even the first three or four chapters are more than enough for you to learn what you want to know about AI.

To get started, you don't need to read the whole thing. So that's a thought. You can read some of the stuff on our blog. You can see what some of the kids have built. There's a lot of, you know.

So we run a research symposium every year. We have kids speak about what they did. It's all like 10 minute talks.

I encourage people to listen to them because hearing them in their own voices of what the kids have done gives you a really good feel for what is this thing? Right.

Speaker C: 00:38:00

Yeah.

Nisha Talagala: 00:38:01

See it from the point of view of a young person and then what are these problems that they're solving and how did they solve them? So that's a good way to, you know, just kind of get started.

Speaker C: 00:38:09

Brilliant.

Joanna Shilton: 00:38:10

Well, I'm going to put links to all of those in the show notes. So thank you, Nisha. And all that's left for me to say is. Nisha Talagala, thank you for coming on Women with AI.

Nisha Talagala: 00:38:19

Thank you very much for having me.