Do you believe that a data-driven business can take advantage of artificial intelligence and you need help taking the first step? Today, Neil Sahota joins us with incredible insight into the world of artificial intelligence. Neil states that AI is not for the future, it is for today! Neil is an IBM Master Inventor, United Nations (UN) Artificial Intelligence (AI) subject matter expert, Faculty at UC Irvine, and author of Own the A.I. Revolution. With 20+ years of business experience, he works with organizations to create next-generation products/solutions powered by artificial intelligence.
In this episode we discuss:
What is Artificial Intelligence?
What is the difference between machine language and artificial intelligence?
Examples and applications of artificial intelligence for small business.
We hope this episode helps you learn about repositioning artificial intelligence for social good.
Josh Fonger (00:24-01:24): Welcome to the work the system podcast where we help entrepreneurs make more and work less using systems. And I’m your host Josh Fonger. Today we have a special guest. We’ve got Neil Sahota who’s going to be talking to us about how to understand and reposition artificial intelligence for social good. Neil is the author of own the AI revolution. His podcast is titled the future when the Assata and can be found on Apple podcast. Neil is also a professor at UC Irvine and a global recognized speaker and author Neil is the founding member of the UN’s AI for social good committee. And he’s here to talk about the ways we can harness the power of AI and use it for social good. Neal’s book provides a future forward look at AI focusing on how businesses can use it to commercialize while doing good in the world. Okay. And Neil, well I’m excited to have you on. Before we get in with the kind of the deep questions, I want you to tell us your, your backstory, how you got into AI and what brought you to this place today.
Neil Sahota (01:25-02:36): Well, I could probably sum it up in one word accident. I never, never planned any of this. I just don’t one of those guys that, you know, when I’m trying to solve a problem rather than just solve the specific problem, try to do it more generally. And so as a result, you know, developed a lot of intellectual property and about 14 years ago, business intelligence was really taking off, you know, the ability to, you know, get all this data and slice, slice get these great reports. A lot of people were saying like, it’s really cool. The machine has given us these great insights. I started thinking to myself, you know, the machine really isn’t telling us anything when we’re doing the work, but could a machine actually do that? And so I went up developing a series of patents around something I called enterprise intelligence, which now we would call it like machine learning. And that was my first foray into AI. I was working for IBM at the time, trying to turn to a product when there was a little project going on called Watson. And so I love getting involved in that and actually helping to build out IBM Watson and more importantly ecosystem around it. You know, giving people the tools to use the technology.
Josh Fonger (02:37-02:48): So for someone like me who knows almost nothing about AI how, what is the proper definition and is that different? Is AI different than machine learning or are they, are they one in the same?
Neil Sahota (02:49-04:54): That’s a good question and a complicated one. The definition of AI is a bit of a moving target. As we advance, we are quick stations are changing but simply put AI is really, I’m a machine that kind of mimics human thinking that he uses his experience. It’s knowledge to try and figure out, you know, an answer or a solution to something we may not have the answer to. So it can do kind of low, low level admin tasks that require some level of cognition. And machine learning is definitely part of AI. I mean, machine learning is the machine I should learn by doing. You have a lot, a lot of data, that’s the fuel and it looks and tries to determine patterns and kind of wires itself. So much like you took a child to the library and asked the kid to read a bunch of books. He would start learning things from the books, right? But just like the child who needs some structure and guidance. So this is the machine. So as the machine is learning, we haven’t worked with human experts so that it can actually focus. It’s learning, I’ll say in the, in a in a proper fashion, meaning that it’s not making wild assumptions or misinterpreting information. So machine learning is a key part of that. The other key part when it comes to AI is the ability to actually understand natural language, which is something that we as people really take for granted. You know, if I told everyone out there, Hey, I’m feeling blue because it’s raining cats and dogs, most people know what I’m talking about. But if I told that to a machine, what does the machine actually think? Right? It’s going to say that Neil is physically the color of blue because small animals are falling from the sky and that doesn’t compute. So with AI it’s not looking at keywords that are looking at like literal meaning is trying to understand context of the conversation is trying to understand slang, idiom jargon. So it’s trying to get the intent of what I’m trying to say. Not just you know, the literal definition.
Josh Fonger (04:55-05:11): Wow, okay. So it’s not just literal language, it’s a intent. Oh, how do you build a machine that can do that? Can it maybe read body language? You know, what, what, what kind of sensors do you need a machine to be able to sense things more than just words.
Neil Sahota (05:12-06:28): So this is a really good point, Josh. We, you know, we all have cameras on all these devices now, so you could actually have that video fend to an AI, they can watch it in real time, but you slept the teach, the AI kinesiology and are people that have done that. Like IBM Watson has learned from a guy that kid has developed, like technology or issues, had a system back in the fifties that could reveal of a person’s lying. So like on your face, there’s over 2000 points alone that revealed a lie, right? The best human can watch, you know, seven to 12 points in real time machine can watch all 2000. And so if you can teach the machine those points, teach the machine what different gestures mean, you know, like tapping your foot, crossing your arms, the machine will actually then understand CIA, as I’m talking with this person and I see their bio language change and mean something, what does it mean? And it gets better with practice. So it’s not perfect right up the box. It can take several iterations and usually takes weeks to train an AI. It never forgets what it’s learned. So it gets better and better. And at some point it reaches that competence level where we’re going, okay, it’s right, like 90% of the time.
Josh Fonger (06:29-06:41): So for all that application police stations could actually play a little AI machine next to them while they’re doing interrogations or interviews and, and be able to tell lie truth, lie truth pretty competently.
Neil Sahota (06:42-07:03): They could, I mean there’s, and I know in law enforcement there’s a lot of talk about using the technology for that very reason as a lie detector, there’s more concern on the legal side. Is, is that valid? Is it admissible? No. How would you contest that? So it’s interesting. We know the technology is there, the comfort level among us as people not quite there yet.
Josh Fonger (07:04-07:19): How about in the investing world? Cause I hear about it about that in terms of machines doing, making the investment choices instead of people who make choices based on emotion and instinct. Is there applications for that as well going on?
Neil Sahota (07:20-08:28): There? There are, I think a lot of the financial services companies are using the technology and that kind of really in their hand cause it’s competitive advantage. But there’s a ETF called AI EEQ that the entire portfolio is actually decided by an AI and they built, and in addition to using like the standard metrics, right. You know, they are trying to hopefully pull out some of this subjectivity out. But ironically, one thing they’re trying to do is the AI measure the irrationality of people, if you will. You know, I know one of the things I’ll tell them in 10 minutes of the market is people are rational, but I think we’ll all agree, we’re not quite so rational at the time. So it’s trying to, is trying to anticipate like investor sentiment. So like when people are they going to be happy, I’m going to be sad and it made me afraid certain news were to come out, will be happen and try and factor that, you know, irrational factor until it’s portfolio decision making. So I have always believed you can’t predict the stock market because I predict the rationale and people, well it looks like these guys may have built an AI that can actually do that.
Josh Fonger (08:29-08:53): Wow. So I’ll have to check that stock out to see how it’s, how it’s doing. I don’t have that fund. So in the case of the business world part of your, your history has to do with social good and being involved with the UN. How does that, how does the intersection of AI and social good, what you know, you know, how does that mesh?
Neil Sahota (08:54-11:01): Oh, AI, like all technology is a tool and it’s all about how we use it. You know, you can use it to create, you can use it to destroy. And I think there’s a lot of opportunities out there for social enterprise, social entrepreneurship where we can do, you know, make money and do good at the same time. And you know, the UN has 17 stainable development goals or STGs, you know, like zero hunger improved gender equality, you know, better access to health care and they want to make these goals a reality by 2035. Unfortunately there’s a bit of a shortfall and making that a reality. The conservative estimate is about $7 trillion a year. High end is like 20 trillion a year and having resources and things to actually make this projection, do these by 2035 technology we’re finding can be a bridge to, to you know, cover some of that gap. And so we’ve already started up through AI for good 116 projects. We’re using the technology to help make the STGs a reality. Like in Africa there’s about one doctor for every 2000 people. And you may not be close, you might be like, you know, 60, 70 kilometers away from a doctor or hospital. How do you improve care in those rural areas? And it’s like one of the things that’s been worked on is like a self-contained tablet. They eye on it that knows some basic like first aid and you know, healthcare so that a villager can take this tablet and someone gets injured or someone falls ill use the tablet, you know, ask questions, use the camera, talk, diagnose and then you know, the ad can help prompt no cases, probably what’s going on and here’s a treatment. Or if it’s really serious, say, Hey, we have to alert somebody, get this person, young helicopter to the nearest hospital. But essentially trying to enable, you know, we’re, you know, people that are not medical experts to be able to provide someone with basic care.
Josh Fonger (11:02-11:27): That’s pretty amazing. So, so bringing it down to just the average entrepreneur what should they be? Should they be excited or say, should I be concerned? Cause obviously they might have 10 employees and so running a hotel or have an online business where they’re doing coaching. Like I am like, how is this going to affect their future in 10 years?
Neil Sahota (11:28-12:27): I think they should be excited. And for very good reason. You know, really have this AI ecosystem and they’re getting people to use the tools and think about 70% of that ecosystem is actually start-ups. It’s entrepreneurs and they’re the ones actually coming in with the more great, innovative or you prefer disruptive ideas. I think entrepreneurs or entrepreneurs, the advantage of trying to look at things differently. You know the big companies for example, they think very much automation. So you know, they have their existing process or system and they’re saying, okay, can we make it faster or cheaper and less errors? I see the entrepreneurs are coming in thinking like, well I have this new tool set with AI. Can I create a new process and new system, a new product? And they’re really unlocking more of the value of AI that way they’re the ones that are really, you know, to be cliche or changing the game.
Josh Fonger (12:28-12:55): So if I’m one of these entrepreneurs who wants to innovate with AI, do I, should I buy, I mean I don’t even know where to begin. It’s probably a bad question. Sure. Should I go buy some artificial intelligence stuff and play around with it and figure out the solutions or, or I mean cause the idea of changing global healthcare, that seems a bit big. But if I think I want to help companies increase their employee satisfaction, is there like an off the shelf AI thing that I can start to play around with.
Neil Sahota (12:56-14:38): The, I mean there, there are, there are a lot of tools available by all the big companies that, you know, they have nice demos and things like that. But if you’re an entrepreneur figuring out how can I use AI, it starts like with everything else, just think about a problem. I mean, what’s the problem you want to solve? You know, and don’t worry so much about the technology itself. Think about kind of that pie in the sky ideal solution and see if there’s an opportunity. And then you know, you, if you’re not a tech expert, you don’t have to be partner up with somebody say, did this capabilities exist? To give you a real simple example, there’s a company called Serono AI. There are start-up started by a therapist and an arrow, linguist and they wanted help, depressed and suicidal teens and there’s like, is there a venue near these people? He told to talk to some communications, well that’s not going to judge them. And so they’re wanting to like, well we have all this AI stuff, could we do something like that? Right? And so they start thinking about it, if I were to build something, what would that be? Right? So not thinking about the technology so much, but what would that be? And then partnering up with some tech experts to say, I want to do this. Can they help me? And then the experts come in and say, well, you can do these three things but not this one thing. Or you know, there’s actually this capability that might help you do something similar to this. But by flushing all that out, they were actually able to build out that type of AI tool to help people. Right. And it’s actually expanded out in that it’s not just a tool to help depressants, suicidal teens. It’s actually like a communication tool. You actually help us all kind of engage better, understand how to speak the language that’s gonna resonate the best with other people.
Josh Fonger (14:39-15:01): So you mentioned that you’re on the, on the board of several companies kind of helping them understand the future of AI and how it can apply. What kind of advice do you give them? Cause I, I’m, most of us here aren’t going to build a, a for you to be on our boards. So what kind of advice do you give them? Do you tell them certain books to read or certain research to be aware of or what kind of advice do you impart?
Neil Sahota (15:02-16:23): Well first I tell them don’t use AI for AI sake. I know that was the hype and all that, but you don’t need to use AI for everything. Right. And you know, especially as entrepreneurs, we have to take advantage of the money and resources that we have essentially as possible right now where the AI makes sense. Great. We should definitely pursue it. And so I always encourage them that one, you know, there’s a lot of great blogs, articles out there like hog the world has some great information, especially for nontechnical people too. I always tell them like, look, you can’t worry about what people are doing or not doing. There’s a whole world of opportunity out there. AI is triggering what they call the fourth industrial revolution. It’s probably going to change 9% of everything that we do professionally. Personally. People are still trying to figure out what those possibilities are and it’s worth investing a little time to say, what are those possibilities? Think about that pain point that’s, you know, you’ve never been able to solve or the opportunity something always wished you could do. It’s worth exploring to see if AI can actually help you do that. And if you’re unclear, there’s a lot of these AI challenges or things going on where, you know, a little bit of crowdsourcing, they can actually help you figure out and say, does AI make sense to use here? Is it feasible?
Josh Fonger (16:24-17:37): Well, here I’m going to ask you a question that I’m thinking about as you guys were talking here just to, just to solve my problem and then I’ve got you for free. So, I work a lot on foreigners, like thousands of them, right? We’ve had millions dollar book and so I get them at a point where they’ve hit a glass ceiling. Either they’re running out of time or I have money, maybe their strategy is bad, maybe they have a bad personal life. Their business systems are bad. There’s like hundreds of variables that have made them stop growth in their business. That growth might’ve stopped at $100,000 revenue, $1 million revenue, $5 million in revenue, but they’ve, they’ve kind of plateaued and I have a lot of inputs about what’s it, you know, I would love to have a machine that can interview all every single client I’ve worked with over the last decade and then come up with like, Hey, this is what we’ve actually learned. There’s 700 variables and these are the ones and here’s the ratios. And you know, I’d be able to tell me that in real time. So when I talked to someone, I’m like, okay, this is the, this is the trigger point for you before we can break through that plateau. And all I have is my experience with coaching. So many clients, but I mean would there be a tool that could do that? Maybe there’s one off the shelf already that’s created.
Neil Sahota (17:38-18:18): I don’t know if there’s one off the shelf. But you can do it. I won’t, I won’t sugar coat it. It won’t, won’t be easy. You have to have the data. But it sounds like if you could go to interview all your clients for the past 10 years, you’ve definitely would have the data, right? You could have the AI sit there and listen or even prompt questions to help flush out. So you get, you know, really good robust data. But data is the fuel for AI. And I think that’s a lot of things. A lot of people don’t understand. It’s not just magic box that I wanted to do this, it’ll just be able to do it. We need data, we need training. And it sounds like, Josh, you probably have both, so you could probably do this.
Josh Fonger (18:19-19:01): This is kind of an interesting idea. We’ll see. I thought you’re going to be a boring interview and now I’m excited. So Oh, that’s pretty cool because if this definitely is a challenge that people can’t solve in real time themselves, but if there’s a machine that’s got, I think you made a good point, the data aspect and humans very easily can get skewed by not enough data or looking at data incorrectly because the emotional aspect and whereas a computer can, can sift through how much data does, does a AI machine really need to start to learn? It doesn’t need like a thousand inputs of data or is it just after 10 it starts to kind of learn from that.
Neil Sahota (19:02-20:34): It depends on what you’re trying to do and how much complexity or variability there is. Like if you’re looking at something that’s kind of very binary, it’s either X or it’s why you don’t, you don’t need that much data. You probably need more than 10, I’ll call it tens record sets, but you probably could get away with 500. Now if there’s more outcomes or more variability, like, Hey, you know what a thousand things can happen, you need a lot more data. Just to put a perspective, Google’s deep minds alpha go. The AI that beat the go champion, the way they trade it is they just gave it the rule book for the game. And that is what we call a ground truth. So it’s rules of how to make decisions. Then they had to watch 10,000 hours of people playing the game. They didn’t say these are good players, bad players, good strategy, bad strategy. Just watch that. And so I’ve learned woos and other tactics by watching, by watching people play. And then they had to play itself a million times. So we call that reinforcement learning. That took about one week for AlphaGo to do. But that gave AlphaGo the opportunity to suddenly test out different strategies, different tactics, all these different things truly master the game. So in that regards, there wasn’t actually, the 10,000 hours of video footage seems like a lot and it is, there’s a lot of variability in how you play the game. But after that it was ready to go and actually start playing against humans.
Josh Fonger (20:35-21:37): That’s just a, it’s amazing. So so for those of us who have a lot of employees, 10 20, 30 years from now, our landscape would be very different based on either their with machine and they’ll have like cyborg helmets on and feeding and information or they’ll actually just be machines entirely. That’s, that’s pretty amazing. Now, I asked you before, before we’ve been chatted today, and this is cause this is just one of my clients and I’m curious he teaches people how to speak Spanish. I think people who speak English just to learn Spanish and very successful. It’s, you know, internet business. And so for him, he, he’s mentioned AI before and you know, someday like what, what are the, you know, how does he know when to cut and run or what, what kind of, what does the future look like? Because the technology might be there like driving cars, but how long does it take till mass market actually accepts a machine as opposed to a human? Is there like a 10 year gap? Once the technology is there, then humans actually are willing to trust it or what, what’s that? What’s that time period?
Neil Sahota (21:38-23:30): No, I to be honest, I think that’s the challenge that we have as people is we think we have a much longer runway there. We’re really, do you know that the only constant life is change? It’s just the change is happening faster and faster. I think when it comes to AI, I don’t think it’s 10 years for a lot of stuff anymore. It might be six, I guess it might be four or less. But you know, language translation for example we’re probably almost the cusp of the change happening that, you know, there’s a lot of tools out there. People are already got like one too many translation. Like if I talk in English, it’s actually some tools out there already where if your native language is French, you hear your French, that person speaks Japanese or here in Japanese vice versa. So I think the adoption is going to be based on the trust of the technology as well as is it good enough yet? And I think we’re not quite at good enough yet but we probably will be in the next couple of years. It’s very similar to self driving cars. You know a lot of people I talked to think that’s 10 15 years away. Singapore has had self driving cars for some, sorry, self driving buses and taxis for two years now. In fact, I know that Audi’s been, we’re working with China. China has one of the most complex setting driving rolls out there. They’re going to achieve level four automation early next year and most likely try and will then certify. So frightened cars to be on the streets. The California legislature is actually currently Katrina bill right now to make self driving vehicles legal on highways. Meaning, you don’t need to have a person behind the steering wheel. And Ford is gonna release a car in 2021 no steering wheel, no brakes, no accelerator.
Josh Fonger (23:31-23:39): Wow. Yeah, that’s a, that’s amazing. So what, what do we do? What can humans do better than machines?
Neil Sahota (23:40-24:53): That’s fun. A lot actually. I know it seems like that’s not the, but when it comes to like the imagination, creativity, I’m doing more really highly complex or things, a lot of variability. We are much better than machines. I mean, don’t get me wrong Rasheen’s they do what we teach them and if we don’t have a way to kind of commoditize something and teach them that they’re not able to do it. And think if you’re going to teach someone to imagine different kinds of life, like silicone-based life, could you teach a machine that? Right. But that’s more a creation. That’s more imaginative. It’s more philosophical, if you will. So I think in the most, like all technology, it’s gonna re reduce the burden on some of the more low level admin type of things. But free our time up to work on more complex work or high value work. Like I look at doctors, you know, if you can help summarize the patient information and help categorize some of these things so that the doctor doesn’t have to do so much paperwork that will free up their time to actually spend, spend time, more time talking with patients. Right. Which we know is really important.
Josh Fonger (24:54-24:56): So you won’t be seeing any AI philosophers anytime soon.
Neil Sahota (24:57-25:04): I, I don’t think so. Right. Somebody out there might prove me wrong in the near future. I don’t know, be in the near future, but probably in the future.
Josh Fonger (25:05-25:17): That’d be a good, good challenge. Oh cool. So what, what did I not ask you that I should have asked you? I mean, do you got a big you know, a lot of knowledge in this particular subject, but what do you think, you know, you want to leave the audience with? I didn’t ask you about,
Neil Sahota (25:18-26:47): I, you know, I, there’s a lot of fear and concern out there, right. And I, and I totally get that and some of it is definitely warranted, but I really just want to emphasize that this is a tool that we use. Not, this is not Terminator time, right? There’s a lot to be hopeful, but we just have to get in the mindset about the opportunity that this, this is a chance to actually not just, you know, do more important work, but in fact probably make ourselves better. A lot of people ask me if AI, this is the century of the machine, and I actually believe it’s the century of the human. They’re just quick example. There’s a project going on called loving AI, or they’re trying to teach a machine unconditional love because the largest, you know, illness is actually loneliness in the world. About 40% of people suffer from that. They’re wanting to quit, you know, a companion. But the question became a trying to teach the machine, well, what’s unconditional love mean? What, what is love and what are different forms of love and why is unconditional love different? The love is turned to this really deep self-reflecting journey. And what it means to be human. That’s become a really deep exploration on us as people and why we might do things or not do things. And as a result of trying to, you know, quantify this for a machine, it’s really put us into a deeper touch with our own. And I think that’s a huge opportunity for all of us.
Josh Fonger (26:48-26:58) : Wow. Well, very interesting. I’m sure we could mind that topic for a long time, but I know you’ve got to run and so do I. So where can people find you if they want more information or they want to check out your book and you know, where should they go?
Neil Sahota (26:59-27:10)They should go to my website, Neil Sahota dot com or they can find me on LinkedIn or, you know, I’m always posting and sharing information and you know, please reach out if you have any questions or you just want to wrap.
Josh Fonger (27:11-27:47) Okay. Very good. All right, Neil, thanks for being such great guests and thank you for tuning in today to the work system podcast. Tune in next week, we’re gonna have another great guests like Neil sharing the future of business, how to improve your business or into one of my past guests talking about how they systemize their company to make more and work less. Also, if you want to get a copy of that book right there behind you, work the system, and if you want it mailed to you, I free edition and leave us a review and we’re pulling the name out of a hat a once a week and mail out a copy of that book. Just send us a picture or screenshot to info at work the system.com and we’ve only got one a week. Otherwise I will catch you all next week. Thanks again Neil.
Neil Sahota (27:48-27:50) ; That’s pleasure Josh!