Human-Centered AI, with Beth Rudden CEO of Bast AI

(upbeat music) - You're listening to Practically Tomorrow, a podcast produced by GoTo that digs into practical, ready to use strategies for making the next big idea work for you today. Each episode breaks down an emerging tech trend in AI, customer experience, security, and more to help your entire organization navigate this fast evolving world. With that, let's jump into Practically Tomorrow right now. Welcome to the Practically Tomorrow podcast. I am Olga Londanova, chief product and technology officer at GoTo. In today's episode, we are diving into the world of Human-centered AI. We'll explore what it takes to build and implement AI-powered solution for high-state business problems. Joining me today is Best Rodin, the CEO of Best AI. Best founded Best AI in 2022, the Deliver Explanable AI solution before launching Best AI, Best Cated Remarkable Carey at IBM, including Chief Data Officer, Chief Data Scientist. This is also an accomplished inventor with most enthusiastic patterns and the co-author of AI for the rest of us. Best, welcome to the podcast. Super excited to have you here, I guess. - Thank you, Summer Tolgat. Super excited to be here. Thank you for having me. The first question that I would love to ask you, so is it just for years? Generative AI has transformed from an experimental technology to a daily assistant, but definitely it forced every person on the planet, so to speak, to pay attention. And so it is how we do the work and it's prevalent in our lives. You have a beautiful background, that blend, anthropology, and data, and AI, and business. So you're bringing to the table a very brilliant perspective of both innovator and scholar who studies human societies and culture and how they develop. So I've had to ask, I would love to hear your perspective of how you see AI impacting our lives, that will work live in Zenets 10 to 20 years. - Okay, I'm gonna do this at three levels, so I'm gonna start with myself. And I am somebody who has been using artificial intelligence in some way, shape or form pretty much since the early 2000s. And one of the things that I learned is that it is very difficult to give AI systems an understanding of the meaning of words. And so when I use AI systems, I'm using it personally to augment myself so that I can go spend more time with my humans. And so I'm using it to do something that is, that is inhumanly, it is not possible for a human to do. And that's where I wanted to start is like for myself, I augment myself every day with all kinds of AI tools that make my life easier. And in the last four years and the last four months, all of these new tools that are coming out, they make somebody like my life better every single day. And it gives me more time to do the things that I value. And so with that, I always say that AI is really great if you're already wise. And what I mean by that is that I already know that if I really wanna understand something, I'm going to have to read and be frustrated and have tension and have friction in order to learn something new. And I know that if it's too easy, it's not something that I have probably done a good enough job of understanding. And that's something that I learned, therefore I am already wise, therefore for me, AI is exceptional. And I do a lot of talks and I even show my work about all of the different AI tools that I use in my talks to either generate pictures, to generate ideas, to generate frames. And I love to do that. And what I've noticed is the tools change so frequently because there's new tools coming up every single day. I'm gonna go one level above and I'm gonna say that if we look at our community and one of the things that I've been interested in lately is when people talk about AI literacy, I of course take it literally and I say, what is literacy? What does that mean? How do we get everybody to be AI literate? And I was thinking about the work that I'm reading is from Mary Ann Woof and she does a lot of work on reading and how humans learn to read. And the book that I'm formulating a lot of these thoughts from is called Proust in the Squid. And what she says, I think is really seminal and it got me thinking about the fact that reading is not a biological evolution, it's a cultural adaptation. It's something that we learned over the last 10,000 years. Every human on earth, every human in a developing country on earth has about six to seven years that they spend during the most, the time that they can learn the most about reading and they learn to read and comprehend the meaning of words. And this is something that AI can't do right now, not effectively. And there's lots of us that have an understanding of why that is. But I want people to know that if AI literacy is like reading literacy, we have a lot to figure out in order to be able to get every human on earth to be AI literate or even business leaders. And so I always try to give that understanding to let people know that they're not behind. And that there is so much to learn, we don't even have an alphabet or phonics for AI yet. And the third sort of tier that I want to go to is my future of what I want for AI. And what I see, and I work with a lot of librarians, and I work with a lot of people who are in positions that actually define, develop, and curate the information for a community. And librarians are doing a lot more than that. Every day, they're interacting with customers and users and people who are in need. And imagine being a librarian where somebody comes in and asks you a question, like, where can I find a book about Brazilian cooking? Is that book in the cooking section, or is that book in the Brazilian section? And we as human beings have positions and roles in cities that curate the information for that community. And so my future is that libraries will be like language models. So you can interoperate with the language model of Detroit or the Bronx or California or some place that has local flavor and local customization. And language models will be community driven, like utilities. And it will be something that everybody uses and can understand more about that local community from that language model. And that's how I choose to see the future of artificial intelligence as a resource. And that resource is a community resource that follows a very similar pattern as utilities do. And if you look at electricians and you want to build a house and you want to put electricity in the house, you would call up an electrician. And if that electrician put in the electricity in your house that didn't follow code, then that electrician would be held accountable by that electrician guild that is local. And this is how I've always thought that data scientists should be held to a code that is understood that they need to make the data available to the domain experts so that the domain expert can understand what to do with that data. But I think we're going to have all kinds of new jobs. And the new jobs and the new roles will be something that has standards for safety, like electrical codes. But I also think that it's going to be much more attuned to what humans need and what humans are good at. And that's intuition and emotional IQ and common sense and creativity, like those are human things. Machines aren't really good at this. And then the machines, the AI systems, they're going to be doing the language-- models and analyzing huge amounts of data and giving you probabilities and sensing pattern. And I think that, you know, that's sort of the future that I see is that you're going to have librarians as stewards for what information goes into that local language model that is a utility. And then you're going to have data scientists or specialists that follow codes for safety and standards, but also are working with people who could then have more time to interact with other humans in a human way. And I think we're looking at especially the only crystal ball in the entire world that any scientist will agree with is demographics. And if you look at the demographics of what we're seeing two day and the demography, we know that our children will be interfacing with more older adults than we ever have in our lifetimes. And this is going to cause people to think differently about aging, about taking care of people, about the needs of the body that fail us regardless of what we do. And so my bet and my view of the future is that you're going to have people who are interested in local culture able to access that from wherever they are in the world, making time for themselves to do human things and connect and build relationships with other human beings that are more personal and probably more aligned to caregiving and understand a way that the data and the information that go into these models and that are trained by these models are created by human beings that are held to some form of a safety standard. So can you tell I've not thought about this at all? Definitely not a passion that's about it at all. I'm also very passionate about a thousand community developed with elder people and younger people and just creating the environment that is conducive to generations writing together. I think it's a beautiful, beautiful model to think about how I can help healthcare and also be critical and for more things. If I think it's more mundane, like work environment, they're also picking up on what you were saying about learning. We see it today everywhere, people who are curious and open minded and they experiment with AI. And I see it at work actually every day. I'm working with some social engineers. That's something that has been transformed by Iatrashur. But you need to know how to use it. You need to spend time to understand how to be good at it and how to partner with this powerful tool because you could, you could program together, right? You could just bounce ideas over each other. But you really need to invest in it. You need to figure out how to do that. And that's part of what we're all going through today. If you said we're just at the beginning of learning Alhobet and we will have a long way to go. We will be getting better, right? And then you're going, judging it beyond, it's like person personality and person that is interaction with it. I definitely can see that the future, you know that yet, but it definitely is definitely coming. So, based on what you just covered, I took from it that you are looking at AI as augmenting human intelligence. AI and humans working together and that's going to give the future because there's a lot of conversation. Will AI take your job? Will AI implant your life? How do you think about it where you are on the spectrum? I often remind myself that, you know, I've been thinking about this and doing this for a really long time and this is my area of specialty. So when I'm giving, you know, I tend to be a little bit too flippant and I want to just take a step back because when we say all AI augment, augment humans and when we put that as a principle for IBM that is still on their website, this is their first, there's three principles and this is their first principle. A lot, you know, you can kind of had different agendas for it. So all AI augments humans. I'm of the practical mindset, which I know you and I share this Olga. If you try to go to a human being and say, can you tell me all of the things that you do in your job so that I can automate that job away? That human being is not, that person will resist. They're not going to adopt the AI. They're not going to tell you all the things that they do in their job and, you know, if anybody has ever worked in workforce transformation, walking up to a person and say, you know, hey, I've got 15 minutes. Could you just run through what you do on a day by day basis for me and just write it all down so that I can shove it into an AI and have an AI do that? That doesn't work like that. So practically, we need to approach AI and human beings in a way where we truly understand what is the human doing, what is the machine doing? And then I'm a big fan of measuring what you treasure or measuring what matters or measuring for insentification of the behaviors you want to see. So if you measure what the human and the machine is doing together and we did studies on us, like tons of them, like it's so much more effective when you give a bounty to a human being and instead of prompt them and say, Hey, I see that you are drowning in paperwork. Could you do some research and figure out if you could do AI to take that piece of your job away? And I'll give you a bonus if you do. You're going to get a whole different aspect in a totally energized workforce who wants to automate the robot out of the human or automate the robotic parts, the repetitive parts, the things that they hate doing, you know, automate that out of what they have to do on a day-by-day basis. And there were, this was like early on back in the day where there were two studies. There was a McKinsey study and an Oxford study. And the McKinsey study said, AI will automate 97% of all jobs. And then the Oxford study came back and said, actually, we broke it down to tasks and AI will only automate 5% of total human jobs. But it will automate all of these tasks for almost 100% of the population. And so when we say all AI augments humans, I want people to go, Oh, okay, well, what human? What is that human doing all day? You know, what parts of their job do we need to automate for them? You know, what do they do that is something that is repetitive, that we don't need them to do anymore because we want to free them up? And I have so many stories of this. My latest one is that I'm working with a hospital where they found that 25% of their nurse's time is spent redirecting family members to something on a website. And so when we implement the AI system, we're going to free those nurses up so that they can spend more time doing therapeutic listening, which is a far more effective way to match what they are good at their zone of genius to the humans that need it and be able to automate those tasks that are easy for an AI to say, Oh, family member, this is what you're looking for. And this is something that we can do over and over and over again. But we need to start with not trying to say we should automate all the cost out of our business and we shouldn't see human beings as cost and we should not be using the word headcount. And you know, so many of these things that we got used to, we need to have humans. And I remember as a very young graduate of a university with an anthropology degree saying, you know what, I really want to put the word human on a slide and being told no that I could not, that I needed to refer to people in the organization as FTE, headcount, everything. So I, of course, totally was a rebel and I used to call them belly buttons. You get the point. So what is a great, great tradition into the next topic. So for different leaders, how do you recommend they identify where AI can truly add value to say, please. I am going to. to tell them the secret of the universe. And this is a secret that only very experienced business people know, and very experienced AI developers who have actually got their AI systems put into production and adopted by users, and a return on their investment for their AI investment. Are you ready? Yes, go for it. The very first thing that all leaders should do is understand the business opportunity. And we did back in IBM, I did a bunch of work with the design team, and we created a series of design thinking workshops where people are actively working together. And these workshops have been battle tested for all businesses across all industries, across all sectors. And this is a really simple thing to do is get a diverse group of people in a room, or a varied group of people in a room. So if you have like a university, do faculty, administration, and students, because many times shockingly, administration creates solutions for faculty and students that they don't use or faculty create solutions that students don't use. So you need all sections of your business to get in a room, or some representative of those sections to get in a room. And you ask three questions. We struggle today because we can't get any more top line grids, for instance, or we can't get more students to enroll. Or there's too much AI hype, and we struggle today because there's too much information. It's moving too fast. And then the second question that you need to ask the group is, we need ways to grow our top line grows, get more students enrolled, figure out what information we actually need. And the third question is the most important. This will achieve our strategic objectives by growing our top line will allow our company to grow, which will allow us to pay our shareholders more. Growing our student enrollment will allow our community college to grow, that will allow us to be able to better deliver education to more people. Understanding what is the important information through all of this hype will allow us to apply AI in a way that really matches what our business objective is. And so those are the three questions. That is the magic sauce. What is the problem? What is the opportunity that your business has that you really want to solve? This is where you can start and start to understand what are the problems that you have today. Start with, we have to put AI on it. Start with what are you struggling with today? Yes, AI might be a solution and it might not be a solution, right? But I am so passionate about it and so glad you brought it up. All my experience with AI, I think during class 10 years or so. And this generative AI, even to a large degree, is about when you build it, was it you build it for your company or you build it for your customers. And for me during at least last six years, it was building for the customers. In order to be successful, you absolutely have to understand what is a problem, what is objective, what are your customers need, what do they want, what are their essential problems that they have. And then figure out where AI can solve this problem or there are easier ways to solve it. And there are easier ways to solve it. That's good, right? The most elegant solution is a simple solution. But there are so much today that you can do with generative AI that we couldn't do before, right? And you start looking at possibilities, then you have to understand generative AI for sure. But you have to match a problem to a solution. And it is generally, if AI models change quickly, right? So you're constantly looking at, is it good enough to solve this particular problem or I need to solve another essential or I can solve another essential problem. And on top of it is what's super important is user experience, right? How we make sure that solutions are not only practical and the problems can be solved, but there are always edge cases that AI is not good at solving today, right? So how do you do it? How do you do in such a way that it is ideally you would call it? You call it practically, you can call it boring AI. You don't want it to be smacking your face and I am AI. Please, let me tell you that I am an AI. You want it to be so natural, especially if you are solving for people who are incredibly busy, right? They're doing the words that is really important. You want it to be natural. You want it to be easy to adapt and feel that you are comfortable with using it. I think that's a super important feel and you have to be doing it all the time with your customers. You have to be experimenting with them. You have to see what works, what doesn't, adjust the experiences because we are learning, we are all learning how the world works and how customers read the work with general APIs, how generality of AI, what generality of AI is good for, are excellent, always and how to handle edge cases. I think it's a fucking-n-a-ton world, I do no longer. I think, you know, and what you're touching upon is the customer experience which relates back to, it should augment the user, the person. So you have to understand what is the experience that they want. And to me, this is the biggest unknown right now with AI, like, you know, as an engineer, I can absolutely, you know, handle everything all the way up to the API hooks, but what that user experience needs to be, you have to know that and you have to understand how do you deliver that experience to the person in a way that reduces the friction. And you know, what you're talking about with like, you know, connecting the dots between what's the business problem. And then, you know, how does that business problem impact my people, my humans, my users, and then what are those users doing all day? We do, you know, we can use the word intent here. And when we did these design thinking sessions, there's actually six intents that if you look at and understand how AI functions, the answer is in ways that can accelerate research and discovery and rich interactions, anticipate and preempt disruptions, recommending with confident scaling expertise and learning, detect liabilities and risk. Those six intents are something that every business needs at one point or another. And if you understand, and I, you know, we, we did this where we said, okay, we need to understand the business opportunity, but Olga, we use the word problem. And one of the things that I like to try to do is make people go back to seeing the problem as the opportunity. Because it truly is the opportunity to apply one of those six AI intents to be able to solve the problem and create, create more top line growth, create more space, create more time, create more things that people, you know, find as a problem that we do have solutions for. But I feel like it's like everyone is just missing the biggest thing, which is like you have to know what is the opportunity in your business, which means that you actually have to talk to the people or in your case, talk to your customers. Is this an opportunity that you can help them solve with what you do as a business? Exactly. I would like to drill down in one more topic. And again, picking up on your expertise, upskilling organizations for AI. This is a focus for many, many business leaders than exactly just today. If you look at all the surveys, my kin, the reports of companies investing in AI, the Killing reports, 58, greater, etc., and AI adoption. World economic forum estimates that one billion people globally will need to be the skilled in the next five years, as AI, the shape, jobs and skillet requirement. At IBM, you would have killed 20,000 people? But it is amazing. So not many people have this kind of experiences. So what advice do you have for you to try to prepare the organization so they can drive in this AI-driven world? - There's a massive opportunity going on right now with all of the AI hype. And that opportunity is for you as a business leader to figure out what problems do you wanna solve? What business opportunities do you wanna have? And probably the biggest lesson that I think most leaders understand is take advantage of any emergency to create certainty for your people. Because that's what good leaders do is they take the unknowns and make them known. And that's just something I needed to say because I think that a lot of leaders themselves are worried but forget they are in a leadership position so they do have a responsibility to create some certainty in whatever way they can. The re-skilling of a human being or the upskilling of a human being for AI, I mean I have, when we did that we had a massive emergency and a massive opportunity which allowed us to have access to loads and loads of data that we would not have normally had access or had a reason to access. And because we were able to combine all of these different data sets we were able to use AI to understand and sense some patterns within that data but only when I showed that data back to the human being or to the individual. And what I mean by that is we had some older HR systems that had unfortunately only, and this is any company that has a large amount of things or a large amount of people or a large amount of whatever it is you tend to classify it into something that you can kind of get your head around. And so you had people classified as a computer operator and what we found is when we joined up that information with some of the deeper information that that computer operator also had a master's degree in information systems but was just working this particular job because they weren't able to communicate something effectively to the manager or didn't have a career path or weren't understanding that they might have been in the wrong group. We gave the information back to the person and said, here's your digital history of IBM. And we see that you're taking all of these Python classes as a computer operator, what are you doing? And the person was like, oh, I want to be a data scientist and we're like, oh, okay, great. Here is a data science professional path. And this is something that seems very simple, right? But it's about taking the model that we all think about and say, okay, well, we have to aggregate and we have to just, you know, could you summarize that for an executive please? Just give me the three top level bullets. When we do that, we're missing or we're summarizing we're not getting the context. And one of the things that I believe very strongly in is AI can absolutely scale expertise in learning and enrich the interactions because it can, at one time, both sense the pattern over large amounts of data and understand how to enrich the interactions to personalize that data for an individual. It can do both. And used in that way, where you're giving the information back to the person and then giving the person the opportunity that they were kind of screening that they wanted in the first place but didn't have the language to do. This was just magical. And this is what, you know, we, there was a notion of like sort of like bringing water to the village, like the people all over the world, they were screaming that they wanted a career, that they wanted a path, that they wanted to upgrade, that they wanted to do the work. And we could see it because we had all of their learning and they were, they were taking classes, they were learning, they were doing this on their own time, they wanted to do better. And Olga, everybody is like, how are you so optimistic? It's super easy. I actually understand humans. People are good. They want to do better. They want to add value. Most of the time, they don't have the language to do it. And so when you're upskilling and re-skilling your community, the best thing that you can do as a leader is to model how you use AI. And to model how you want them to use AI, to measure what you treasure, to measure what matters. And the biggest reason that we got, where we got in 16 months, over 23,000 humans, is we put a leaderboard out that said, this first line manager has five certified data scientists. This second line manager has 25. This third line manager has 50. This fourth line manager has 175. And it became a race. Measure what you want to see more of as a leader. And that means doing it, modeling it. We had leaders taking classes in data science and learning how to code, which I do not think that we need to do. But they were also taking classes in mindfulness and learning how to fail in a way that people could begin again, so that they could get done with the project that they knew was not adding any value. Have you ever encountered leaders that just don't, they're just stuck, they can't fail or they can't move past something? - Well, it's happened. We have human. - Yeah, it's happened. - The two things that you said, I think are so important for all of us. Remember, model the behavior that you want. And I think in the large organization, it's not going to be one person, right? It stands with the leader and then you have to have more leaders modeling it with you and creating this community or trailblazers, right? And humans learn from humans. And the second one, it resonates with me, because I've seen it with every successful skill in business information or whatever measure, measure what matters. Have the scoreboard and the outcome is so excited just as so many people growing, learning, succeeding, it's amazing. And I think that's what we all want for our organization. - Humans are really good at figuring out social norms. So we as leaders have to make sure that those social norms match what we want them to do. And it really is that simple. And it's a combination of so many different things, but I do think at the core, you nailed it. It's like those, there's two, it's like that. There is something about giving people the incentive in a way that activates healthy collaboration and competition that also allows people to say, you know what, this isn't working. We need to do something differently in a way that I hate the word failure. Like I think that that's not the right word, but it's more like continuous improvement, continuous testing. That's what AI needs always, because it's growing so fast, we're learning so fast, we're adapting so fast. So it's not ever done. And you know, this goes back to the way that I see the future of AI too. One of the biggest things, and I don't want to totally open this up, but one of the biggest things that I see people getting wrong, Olga, data is a flow. It's not an event. It changes all the time. So if you have a model that is not continuously being trained in an appropriate way and continuously evaluating its feedback, you are gonna keep up. - This is beautiful, it's sad. Unfortunately, we run out of time. Thank you so much for joining us. Is it one take away from today? AI is about people who use it, the problems, business problems, it's all. I would live everybody with this thought, learning is never done. And AI keeps changing our world every day. Thank you so much. - If you enjoyed this episode and would like even more practical insights and tactics for navigating the latest tech trends, be sure to hit that subscribe button. Practically tomorrow is produced by GoTo, cloud communications and IT solutions that power a world of work without limits. To learn more about GoTo, visit us at GoTo.com.