Nathalie Post  
In this episode, I am very excited to be speaking to Sudha Jamthe, who is joining us from Silicon Valley where she teaches autonomous vehicles and artificial intelligence at Stanford. Also, she has written a number of books, the last one being AIX: designing artificial intelligence, I have her book right here. And I highly recommend you to pick it up and give it a read. It is full of practical insights, and case studies. And it is great. Now in this episode, we'll be diving into exactly that the topic of AIX and what design means in the context of artificial intelligence. Now, we're recording this episode in early June 2020. And for context, we wanted to start by addressing everything that is currently going on in the world. And all the very much needed conversations that we're having around race and racism, inequality, white privilege, and the Black Lives Matter movement. And also the role that technology plays in this global environment. We believe that anyone with a voice should use that voice, and speak up.

Sudha, thank you for joining me here today. And I wanted to start by asking you, also relating this to the topic of the podcast, humanity centred artificial intelligence. Do you think that data and AI can play a positive role, or could they be used to do more to support the Black Lives Matter movement?

Sudha Jamthe  
Thank you for having me. It's been it's a it's a pleasure to be here. And I'm so glad you started with that topic. Because I think all of us have to do our part. And we are all searching and saying, what are we doing to help our black colleagues who are being marginalised? How are we contributing, or even our silences is not acceptable. So one thing about AI, like every other technology is it can be useful, good and bad. So the way it is used, so the whole concept of AI is you feed a whole bunch of data, and the computer makes patterns and helps make some recommendation, right, so it could be a self driving car. So there's a lot of data that is out there in terms of education, crime level, opportunities, housing crisis, every data is available out there. So I have seen some research, before all the black lives matter protests happened, where some researcher took data where there are trees in a neighbourhood anyways, they said they took when there are trees, and all our trees, in fact, are numbered and with that code and need to add them at the city level here. So they took that data and they found out that the neighbourhoods which had more city more trees, versus the neighbourhood which had less trees, and they it turned out a cities that had, the part of the city that had more more trees was developed better. And that was predominantly white neighbourhood. That the neighbourhood which happened to have less trees, which looks like it's completely not causal from a stats perspective, but turned out that was more black neighbourhood, and they were poorer neighbourhood. And so it goes back to the history in us where the neighbourhoods were segregated, there was this concept of redlining. And they will develop like that. So there was, there is today when I have colleagues who are brilliant people, and we forget our race and everything where we come from, and we have a lot of immigrants right around us. So we forget that when we go into tech, and we start working, but the person who comes from a black neighbourhood who goes to a black school, who's whose house pricing is less, has less opportunity has to fight harder to get to the same level in college, and to the same level to get the jobs. So by exposing this kind of data in research is one thing that AI can help to see. Now we want to fix the fundamental system and inequality and where all is happening, so one is just showing the neighbourhood, the school system, what needs funding, what needs better support. So that is understand just the trees, but you can take all the city data, and you can figure it out from there, right. So that is one that is beginning to happen at the research level, it needs to be adopted at, at the city level government level to make it happen. So that is one other than that, today, when there is this protest with Black Lives Matters. There are, again, so much data out there, which can actually show even the number of arrests that's happening, you know, whether, obviously, it's you know, disproportionately more African Americans are the ones who are impacted. So all that data is going to be available. And data can give us the tools to do something positive with it, and we are ready for the change. So that's a good thing, right there is going to help us. The other side, where AI is not yet helping us is in when one thing is the facial recognition technology is being used for surveillance, not just in us, but I heard in London streets, they have all the traffic lights that are using in Australia, they have traffic lights, which are looking into people's car, to see whether they are texting and driving and automatically sending a ticket. So all of these surveillance technologies, which is powered by AI, is always put out there with a positive views. But they all can be misused. So we need laws in place to curtail them. So we had this thing literally yesterday or day before in California, about this facial recognition technology that was coming up for approval by at the city level in California. And because we are very aware of Black Lives Matters and whatnot, and what's happening out there, people basically call them said you cannot even let this surveillance tools out there without any check and balance. So that's kind of being stopped. And then we have security cameras outside most of our houses. And there is one from Amazon called ring. And again, that could be used and that ring has a relationship with with the police, and they want to monitor our neighbourhoods. And earlier it was present at saying hey, we'll keep your neighbourhood crime free. So today, again, they could misuse that. So there is awareness to that. And people are making noise about stopping such surveillance or misuse of the technology. So that's again, you know, it could be done in a bad way. But because of their weirdness, they're going to stop AI from being evil. And the third and most important piece is all AI is powered by data, which means whoever is developing the AI says, here's a frame of videos or pictures or statistics numbers, right that we are failing to train the AI. And all that AI is bias, because we are seeing this whole Black Lives Matters movement is needed, because there is systemic bias, unconscious bias in society. So that is propagated in AI. So we all of us who are watching listening to this, we have the responsibility to make sure that we bring in data that covers enough diversity, not only black people, but all kinds of diversity, there's not many enough women in the data also. So women is going to be stopped with the false positive compared to a male, right. So and there is this whole bias against LGBTQ people. So all of that making sure that data is unbiased, and has enough diverse data is very, very critical for AI to be functional to reflect the society that we want it to be. And the only way you can do that is to give opportunity for enough black people to be doing this work. And there are brilliant people, it's just because of the systemic bias. They may not go to a school, or they may not be in the network from where they could get hired. So I think each of us have to do our part to make sure that we look at our colleagues and look at how many people around us and whether you know whether we have equal representation. I was doing some soul searching and looking at my students I have taught in Stanford, I teach AI and autonomous vehicle business at Stanford Continuing Studies and their professional adults who come from work, take the extra time to go learn this new exponential technologies and be with their carriers. And I'm doing some soul searching and I don't have 50% black people in my in my student population. And I am going to figure out how I'm going to rectify that. So I think each of us have been do our part and then AI will do its job and do good for us.

Nathalie Post  
Yeah, no, I fully agree with what you said. I think this is for all of us or for many of us a moment of thinking, how can we do more than what we're currently doing? And how can we educate ourselves better?

Sudha Jamthe  
there is one site called Black who design. The whole site is called Black who design. And it's full of people who are, you know, from the African American community. Or it's, I would say that because it's global. So I would say it's black people from everywhere, who are founders, entrepreneurs, UX designers, everything. So I think if people who want to actually go and hire people, so it's called Black who dot design. So it's a curated list. And you can go and it's very inspiring for me, because there are freelancers and graphic designers and, and everything, writers, you know, the whole whole spectrum, but it's a it's a tech lens, right? in tech industry, everybody from different jobs.

Nathalie Post  
And so I mean, talking about designing artificial intelligence, your book, AIX, obviously, is hitting on a lot of those points already. But could you maybe talk a little bit about what AIX actually is, and why we needed and I think a lot of what you said emphasises why we needed in a sense already. But I'm very curious to hear your explanation.

Sudha Jamthe  
So AIX is a word I'd coined for designing with artificial intelligence. So today, we have AI everywhere in technology companies, it's not just the product that they're releasing that has AI, they're actually using AI a lot more internally. So many of the way design was done in the past. And I've had, you know, the privilege of working with amazing designers over the years. The way design is done is you do user research, there's a design process, right, you do user research, and then you actually say this is the customer problem I want to solve. And this is the design experience I'm going to give them so either to design thinking or to service design, you go through the flow, and then you actually create the the visual UX of the UI of it, right. And the way it is done is using storyboard and then you iterate as a team to figure out that was the right thing. So now what has happened is everything is automated, that is not of data feeds, and everything is dynamically ascending as a screen. So the engineer is coding and pulls this data feed and create some experience. And that experience is not going to be perfect because it does, it's not done with a design eye. And so one example I can give us in the car, because I operate in the ad space. And I research that there is so much data in the car. And there is so much opportunity of creating useful experiences with the car, we are still driving while we are driving before we get to autonomous vehicle, we are driving the car, like a horse buggy. We are everybody there are four people in the car and everybody is looking up to people at the back, I'm looking at the back head of people in the back. That's not good design. And it's not a good experience when you know imagine a family is going on a long trip and the kids are screaming. And the the person one person is driving and the other person is supporting them and then turning back looking at two people at the back. And that is not a it is not the ideal experience. So there is a lot of opportunity to fix that experience. But because the data feeds are coming in, which will help create maybe entertainment which will help understand the context of what the people are. So designers coming back to AIX. AIX is about designing with AI. So design and AI talks, the only language AI understands is data. So for how do you design augmented reality experience or voice experience? Or it could be pure machine learning feeds and then you're creating a ecommerce recommendation experience. All of that is going to be coming as data feed. And how do you create that experience is what AIX is about. And I can take Voice as an example because from a consumer perspective, it's easy to understand that so we have voice assistents which are fun with that we just chat with or play games or check weather or do something mundane right. But cars are increasingly bringing voice as an interface inside the vehicle. And with AI. It is not just voice is one interface. You're going to talk the car is just going to be driving. So when everything becomes autonomous which is what it is tending towards. So the car when it is talking to you, you think the car is the AI. So if you take Alexa, some people have taken Alexa and they brought it to the car. Amazon also has that as Alexa option in cars, in certain one race car brand. But Mercedes has a native voice interface, you can say, hey, Mercedes, and then even talk to the car and and ask for stuff, you know, then you're interacting with the car. So as a designer, you have an interaction with the product, right? It's not a computer screen, there is no context. What is the personality of that? So in designing for AI, the AIX for voice, going back to the design process now needs to think about what is the user persona we are we are designing for, and what is the AI persona, because is the AI going to be a responsible, mature adult is going or is going to be a quirky little thing. It's going to be funny how that is now left to the designer. And my whole motivation to write this AIX is if you do not give, bring the designer and educate them, we cannot get this data out. We just cannot do it. We're stuck in the AV space. Because we don't have enough designers creating the in car experience and the intricate experience. Yeah, the second piece of AI x is the product manager. So the product manager is like the CEO of the customer. And they said this is the customer problem we're going to solve we're going to create the the entire design experience not there's not just the UI of that right. I get product managers need to be involved to know how to do this with data.

Nathalie Post  
Yeah, no, exactly. And so how do you see AIX happening in practice? So I know you're talking to a lot of organisations, very different people. Do you see it being adopted? Or or not?

Sudha Jamthe  
So since I coined the word AIX, I think it's not called AIX anywhere. So I'm actually seeing one of the things I always do is I know I'm a technology futurist. So I think of where there is a gap to innovate. And then I go and talk to people in the industry and say, Is this real? Is this I'm kind of making things up? And is there business value in it? Is there a problem? Why is this not done? Or if I tell them about this, are they excited to you know, run with it. So I have talked to a lot of companies. And the closest to where the AIX is being used today is in data feeds from search, the search experience is actually computer powered. So it's an AIX experience. A lot of commerce happens with AIX. And so one of the things I do do have in my book is I have 10 different case studies of companies from around the globe. And I'm very, very particular since I come from Silicon Valley. And it represents tech, I don't want it to be Silicon Valley talking because customers are everywhere. So I've done case studies from around the globe, showcasing their AIX is used. And one of them is this a bunch of chat. Anytime you go to a website these days, you have a chatbot. And those chat bots are AI, there is one company called UIB, who is actually partner partnering or instead supporting the big telecom company in Dubai, to manage their customer experience customer service where people may need customer service call, instead of call, they could do a chat. But it's not just a chatbot chatting and kind of leaving the customer hanging. They've done some kind of AI intelligence on the back end behind the chatbot, which ties everything together. And they've done it in a very measurable way because the telecom company is very particular about keeping the customer happy, and keeping them engaged. And not just reducing the call volume. So they've done so in customer service AI is there and AIX design is being used. Big time it may not be called AIX all the time. In terms of machine learning where it is invisible data feeds that has been used. The other day, I heard about a combination of these, right. Where it's not just one AI because it's not just about AI, it's about solving a problem. Technology is a tool. So I heard about a hospital, which is using a voice interface for kid but kids in ICU when when they leave the kids, and they need to be called and the kid might not be able to call or communicate like an adult. And so they have a little teddy bear, which the kid can hold or talk to. And based on listening to their voice and tone of voice. They are actually trying to use AI to figure out whether it is it's distress or it's a happy chatter or whatever you know and then then based on that, it will, it will trigger an interface to call the care provider or to call the mom or you know, something like that. So it's basically, if you skim down this technology, this kind of voice interface with a AI expert voice, which is attached to machine learning, which does the rest of the calling in the calling the adult piece, but it needs a very sensitive AIX experience, which is which exists in the hospital setting. So I see where there are consumers like we have today, we're talking about facial recognition and surveillance software, when you deploy them in airports, you cannot get customer buy in you cannot make this move, if it's not done with design with the right design thinking.

Nathalie Post  
And where do you think that AI can make such a positive impact in a good way in a way that is good for people that are affected by it or impacted by it in some way.

Sudha Jamthe  
So the beauty of AI or the scary part is that it's pervasive, it's everywhere. So in every walk of life, the easiest example we see is in our homes. You know, it can you can, you know, cook for me, I have this thing called rotomatic, which is a robot which makes my duties that's all fun. And we are early adopters who are playing with Alexa or this, I don't believe home is going to be the first front where we are going to see AI. So I since I am biassed a little bit because I research autonomous vehicle space, I see a lot of AI in connected cars, which are becoming autonomous. So many of the car companies are using AI big time to just reduce cybersecurity because they're connected now to to improve cybersecurity. They are using something called healing AI, which is where the car is running. There's so much technology in the car in any ordinary car, not just acknowledging the car, right your power steering, and your infotainment system. And there's a lot of different technology providers will make all the parts they Peltier on providers, which comes together. So there is an underlying technology in there, which is, which doesn't look fancy, because it doesn't have a UI that is talking to us yet. But there could be some hacking that could happen because it's connected. So the healing AI makes sure in real time that there's no unforeseen situations coming up and makes prediction, then the self driving car is one that has been piloted around for many years. With the COVID situation, everything came to a halt. And now they're all having a rebirth. So just last week, I saw four different companies who are taking an autonomous vehicle. And I'm sure the designers have landed because they are creating a good experience to this. And they are deploying it for delivery. So there is one hospital here called Mayo Clinic. And what they have done is they've partnered with the self driving car company, which is more like a van autonomous man, and from a company called navient. And what they've done is they actually take the they have like drive thru stations for getting COVID testing done. So they don't want you to transport that to the hospital to the lab. But so this, this is actually left in an autonomous vehicle. And from then it brings it back all the way to the lab. And then somebody picks it up. So there is a human touch point when they get in boxes. And they see human touch point when they remove it. And that's all right to the car, that delivers and brings it. So that is one way more here just said they want to do some kind of nonprofit art supply delivery. With is kindd of dipping the toe in the delivery world. There's a whole bunch of delivery bots, which are delivering food and stuff. Yeah. So that's the new thing. So that's the UX of that is happening. So the biggest area where AI is useful and and going to be more useful is healthcare. All the you covid tests when they say they're testing and they said it's positive, it's not negative, that's an AI. Just the way it is done is not done manually, right. It is done by AI making a prediction. That's why you see, or this test is not good enough. And it has false positive false negatives, and then people are sent home and they get it and they come back again, unfortunately, and that is all it is. So with COVID testing is a big one. It's being used today, self driving cars, they coming autonomous vehicles in various industries is another one a lot in shopping and anyway that is data. Yes, they are beginning to see that. Companies are not making so much noise about the use of it. Should be used in airports, it should be in malls should be used in all kinds of places that it's I would say that we are staying away from those today, unfortunately, slowly losing our way back. But with this human interaction, which we now know the value of thanks to COVID we need AI.

Nathalie Post  
And also, I mean, what you're mentioning really hits on the point that there are so many areas where we will see AI that we're currently not seeing AI yet or starting to see. But that also creates a lot of new job opportunities, I think. And well, in your book, you wrote a great chapter on pivoting your career to AI, can you tell us a bit more about that and share that?

Sudha Jamthe  
Thank you for asking that. Because in all my books, I have this one chapter on jobs, because especially coming from a technology industry and been there forever, I get very excited by the latest toy. So I have trained myself to calm down, focus on what I believe is going to take off and technology comes up and down in waves. So something might look like quite today, but then it might crash and go away. And you know, Gartner has this thing called hype curve. So I seem to be always like, each year when you catch me, I seem to be on the graph of the hype curve. So I wrote my book about Internet of Things and started teaching that topic at Stanford Continuing Studies, when IoT was at the top of the hype 2015. And then I moved to autonomous vehicles, and the self driving car was the next one in the hype. And now I'm on the AI bandwagon. And I'm kind of riding that wave before it happens to be at the you know, so my my one metric of whether I would bet on that being the next wave that I believe is not going to crash, but it's going to take off his jobs. So you could create all the technology you want, I have brilliant minds all around me. And they will, they will create technology, I come from an engineering background. So I worked in that world of building technology. First, right? Now I know it technology has to solve the problem. And you're not going to go from here, I make technology. And here it's going to solve a problem, there is a huge spectrum of roles and people needed to go back to the designer is one, right you need product managers is a very important role. In fact, all these technologies involve AI have some kind of sensors from where they're getting data. So you need an ecosystem of partners to come together to make a solution for a customer. And so there are various different jobs, even that somebody was doing the market development of how this AI could be used in airports, or self driving cars, or somebody who is basically single escape creator ecosystem of partners, has to create a deal where they are sharing data with GDPR compliance, and doing this right, creating the responsible interface as a responsibile AI. So all of those jobs for business development, market development, technology, innovation, Product Management, UX, all of those jobs have to be cleaned rescale from people who already know to do this in another world to come. And that is when that industry will take off. And I'm seeing that with autonomous vehicles. Now. I'm seeing that with AI. And I'm actually seeing that coming together in some way. So that is my motivation to write the chapter. And when I talk about it as pivoting your career, I'm talking about people not waiting for the jobs to show up in a job rack in a company site. I'm talking a lot. My style is in all my books, I write about the technology, you can go read about this anywhere. And there's tonnes of webinars these days when especially when we're sitting in lockdown. The learning as an adult has to be done with unlearning. So you need to unlearn the old way of thinking, to learn the new way of of the new technology. And then I always help my students and my readers figure out where are the gaps where they could innovate. So I'm not going to say something very prescriptive for you to just say, okay, A, B, and C. And now you're going to go to this job. So what I do is to say, Okay, if you come from retail industry, and you're looking at the green card, where do you think is the opportunity and I get them thinking and give them pointers to lead them in that direction, come from healthcare. So I have students, I had a student from health care who came and sat in my AV class. And she said, I don't know I, I've always felt, you know, I learned something. When I go to a different environment, I want to see how this will be useful. And now we have actually, you know, AV's in healthcare, and after this COVID situation, and the current situation has accelerated the adoption, but after it settles down, they have this deal with this AV company, and they're going to say, hey, does it make sense to deliver using the AV? How are we going to monetize it? That is what it I want people to think about, understand the ecosystem understand various different job roles, how to innovate, how to make this happen in the industry. So in fact, if I, if I can say about my online classes if that's okay, So other than teaching at Stanford, I have a platform called driverless world school.com where I have a series of classes, and it's not huge number of classes, but I'm very focused on a small class sizes, and getting people to go through this journey to pivot their career. So I am bragging now, but I am very proud of my students, they are in every possible job in every possible industry, and a lot of them are in Bay in, in Silicon Valley around me. But I have people in London, I have people in Germany, in Madrid. So the path to AI is my newest offering, which is basically a series of online courses, where I, and i'm not i when i say i, it's not just me, I have a team now of people who are brilliant passionate about AV in different forms. There are people from the car companies who are partnering with me to develop the content. So it's baked in the reality of what applies for different different brands, how they're using that because it may not be the same across time. So I have this one class called AV master class, where people can get the foundation of this whole AV space and expand that how designs to think about how autonomous vehicle will help them, their industry, and I challenge any industry, I can tell you how he is going to impact them by combining with AI. The second thing is business of AV data, which is my flagship course, there we look at connected cars, how to create digital twins, and then how to create experiences is the the final piece. So these are the two courses that I'm offering right now as part to me, then I have an AIX class, which is wrapped up one semester of AIX class, I don't have it right away, but I will be offering one more, very soon in late summer. So the whole above for the support path, right. So at least three or four people need to develop confidence. While they're doing it. I'm calling it paths to AV, because I'm going to give them coaching and help them figure out a plan based on what background they come what skills they come what they need to unlearn what are the gaps, then I bring speakers each week for the class to have a closed room. Q&A. And the real deal industry. Like you were asking all this is good, but who's really using it? So and you talk to the people who are using it? What is the challenge is is it there where you know, it works for everybody, or the way you're, you're thinking about it is that too futuristic is superior, right? So just temper that expectation. So the learning, the networking, the coaching, to have a one on one plan, and then at the end of it. By the end of the year, I would say by somewhere mid fall, I should be able to get a capstone project for students, which will be the final course in the path series, where they work with a real company with real data, get real experience. And so they can move and pivot that carry us to the job. So that's what I call a scarier pivot. The same thing with AI. So if somebody wants to do AI, they can actually take my course and still say, Okay, now I learned the data in the car is is I'm focusing on self driving car as the main character car self driving car, but it ties back to other industries. That's an AI course. Yeah. So they could take that they can pivot into the AI world. So that's the that's how I think I'm and these are the people who are going to make the future as fighting for

Nathalie Post  
Yeah, thank you so much for for sharing that. I also think it's like a great call to like, I mean, call to action to pick up your book, or attend a course. Yeah, definitely recommending the book. I really enjoyed reading it and some really interesting takeaways. And like you said, the case studies are, I mean, I think that is what is missing in some of the AI books out there, where you really want to see it practical, like how do companies do this in practice, and what are the challenges they're facing? So I really love that. So are there any other things you would like to say before we wrap this up?

Sudha Jamthe  
Thank you In closing, I received, first of all thank you for having this conversation with me. I think. I don't usually get to talk to a podcaster who's actually my reader too. So it was a pleasant surprise. The one thing I would say to people who are watching this is, it doesn't matter what industry you are in, it doesn't matter if you don't know coding or Python or you know, you're not deep in tech, you have to get into this AI bandwagon. And you don't have to take a pay cut. And you don't have to say I'm new, and I need to start, that is my biggest advice. You have to invest the time to unlearn and relearn. So we go through this exercise of skill development, saying, okay, what do you bring to the table, it's not the job title that you're sitting with, there going to be new job titles. So but we are we if we are doing the same job for a while in a certain industry, we get why I think that's what we can do. So if you come from retail, into autonomous vehicle, or you come even from automotive, but you're going to go help the telecom sector or anything, right, there's a lot, a lot of industries are being impacted by this called mobility. So I have one recent article in automated buildings, about how autonomous vehicles and automated buildings are on a coalition path. I've written about this while back inspired by the by Ken Sinclair, who's the editor of this magazine, but in recent times, think about this with COVID. We were all moving in cars, or public transport, we were mobile, and we went from home to work. Now we are all forced to stay home, which means instead of the car, we're all confined to the building. Right? So there is a coalition but they I think, buildings and cars are coming together. So I'm giving that as an example that you would never know what industry collides or disrupt another industry. So just don't limit yourself. Don't think you don't know take don't think you're not done. It doesn't matter. There is a role for you. Just get involved. And then I have a bunch of free webinars and classes if people want to just try it out to test that so that you can ping and talk to me. I love helping people and think you're helping them figure it out. If not, you know given for my course. I want only a small set of people who are the right fit with the right timing to make the carrier change within the year style and they take my class. Otherwise, you can take any of my free courses and webinars. I do one on Wednesdays on YouTube, so I'm happy to help people.