Powered by RedCircle
Part 2 is now out!
Let’s continue the conversation with expert Paul Zikopoulos, the Vice President of IBM Skills Vitality and Enablement to explore AI fundamentals, job elimination or creation, using the cloud for storage, and the future of AI.
Paul is an award-winning professional writer & speaker who’s been consulted on AI & Big Data by the popular TV show “60 Minutes”. Paul’s been named to dozens of global Experts to Follow & Influencers lists, including Analytics Insight’s Top 100 Global AI & Big Data Influencers. Paul’s written 21 books – including The AI Ladder, Cloud without Compromise, and 3 ‘for Dummies’ titles – and over 360 articles during his accidental 28-year career as a data nerd.
At IBM, Paul leads from the front, helping to shape the strategic direction in a ‘tech years are like dog years’ world for the IBM Technology Unit’s (all IBM software & hardware) sales, tech sales, and partner ecosystem learning journeys and upskilling programs.
Paul actively supports Women in Technology and is a seated board member for Women 2.0 – now called Switch which he became involved with after his tweet was mentioned on the TV show, The View. He’s the only and first male to win IBM Canada’s Women in Technology Ally of the Year award, and is at the forefront of general workplace inclusivity, completing an intensive D&I certificate at Cornell University. Lastly, Paul is a seated board member of Coding for Veterans and sits on the world-recognized Masters of Management Analytics & AI program boards at Canada’s prestigious Queen’s University.
Conquer Local is presented by Vendasta. We have proudly served 5.5+ million local businesses through 60,000+ channel partners, agencies, and enterprise-level organizations. Learn more about Vendasta, and we can help your organization or learn more about Vendasta’s Affiliate Program and how our listeners (like yourself) make up to $10,000 off referrals.
Are you an entrepreneur, salesperson, or marketer? Then, keep the learning going in the Conquer Local Academy.
The Future of Artificial Intelligence
George: This is the Conquer Local Podcast, a show about billion-dollar sales leaders, marketers leading local economic growth, and entrepreneurs that have created their dream organizations. They want to share their secrets, giving you the distilled version of their extraordinary feats. Our hope is, with the tangible takeaways from each episode, you can rewire, rework, and reimagine your business. I’m George Leith, and on this episode, we welcome back Paul Zikopoulos, for part two of Artificial Intelligence. Paul is the vice president of IBM Technology Group Skills and Enablement. He’s an award-winning professional writer and speaker. He’s been consulted on the topic of AI and big data by the popular TV show, 60 Minutes, and he’s been named to dozens of global, experts to follow and influencer lists, including Analytics Insight, the top 100 global AI and big data influencers. Paul’s written 21 books, including the AI Ladder, Cloud Without Compromise, and 3 ‘for Dummies’ titles, with over 360 articles during his accidental 28-year career as a data nerd. At IBM, Paul leads from the front, helping to shape the strategic direction in tech years are like dog years world for IBM technology units, all software and hardware sales, tech sales, and partner ecosystems, you can find Paul helping improve learning journeys and up-skilling programs. Paul Zikopoulos coming up next on the Conquer Local Podcast.
George: He’s back, Paul Zikopoulos, two episodes, all around AI. It’s one of the hottest buzzwords out there. Lots of people are trying to understand more about it. Paul, thanks for coming back on the show. You did a great job last week of setting the stage.
Paul: It’s great to be back.
Job Elimination vs. Job Creation
George: We promised our audience we were gonna talk about one of the questions that comes up all the time, and I get it when I’m working with sales teams and they’re like, “Is my job safe? Are we gonna automate sales?” I’ve heard from organizations, “We’re gonna automate this.” And I’ve been in a tech company for 10 years, we’re no closer to automating sales now than we were 10 years ago. In fact, I think we might actually be further away because there’s so much going on. But do people need to be worried about their jobs when it comes to AI taking ’em away?
Paul: Yeah, listen. I get this question all the time, especially from politicians, by the way. And I think it’s best to explore it in kind of like an onion and peel the layers. So I think overall I’ll just state this, and I’m not a fortune teller. So I guess I can’t tell you, will AI have a net job creation or job elimination? I actually believe that AI will create new jobs and more jobs than it takes away. There will be some displacement. Look at people who truck the ice around the houses were displaced when we created the refrigerator, so that kind of completely changed everything. So we go through these epochs of technical or some kind of evolution in our society, and it displaces certain work but creates other work. So I think AI is gonna be a net job creator. I think if you ask, “Am I gonna lose my job?” I’d share with you two kinds of anecdotes or thoughts in my head. One is, while I think AI will be a net job creator, I think salespeople, HR people, managers, I think anyone who gets more comfortable with technology and AI, will replace those who do not get more comfortable with technology and AI. And I’m very confident in that statement. So roughnecks working on oil rigs, they’re gonna be the exact same people, but I’m gonna call ’em rough techs working on oil rigs. They’re gonna get comfortable with holding a portable ruggedized device that does visual inspection with AI on the camera and interact with that. And I think that that’s kind of the key thing to think about. So when I think about the types of jobs that we’re gonna have, we’re so used to saying, “Here’s a blue-collar job. Here’s a white-collar job.” But quite frankly, technology will touch all jobs. So I actually call them new-collar jobs. No more blue, no more white. They’re new-collar jobs. And if you prepare yourself to be comfortable with AI, there’s no reason why you shouldn’t. You’re using AI on your phone when you unlock your iPhone by looking at it, that’s AI. So when you start to think about how this will come into your workplace and you embrace it, this doesn’t mean you have to go and create convolution on neural network algorithms. That’s not what we’re talking about. It’ll be a premium paid for those roles, on people that are able to do that kind of stuff. But everyone should get comfortable with technology in their lives. And the analogy I will give you is chess. So if you remember Garry Kasparov, or I dunno how old everyone is that listens to this. But years ago, this was the world’s grand master, the best chess player, and an IBM computer beat him a number of years ago. And afterwards, Time Magazine put him on the cover and said, “Is this the end of humankind and the end of chess?” And many people felt that Garry Kasparov was sour grapes, ’cause he accused the computer of cheating and so on and so forth. But what he is really saying, is, “If I had access to the millions of computer games that this computer has observed, I think I could have beat it.” And so in the years that passed, we have this new kind of chess movement, and chess is becoming pretty cool. Queen’s Gambit on Netflix, there’s this cheating scandal now going on. So how is this all attention coming there? Since that time has been the invention of freestyle chess. And freestyle chess is this chess consortium where you can go and play as an individual. So I Paul Zikopoulos will just play alone, I’m great at chess. I can enter a computer to play. So lots of universities build AI chess computers and submit them just to computers. Or I could be what’s called a centaur chess player, where it’s humankind with machines. So I come in with my computer and I play you in chess, individual computer, another centaur player. Since freestyle chess has come, there’s been record involvement in the game of chess, there’s been record amount of grand masters crowned, which means that mankind with technology can achieve more than just mankind. And if you look at the winnings, this is the most interesting part, about 40% of the time these computers win on their own, 60% of the time a freestyle centaur chess player wins. So humankind with machine. You know who never wins? A human on their own. So I think in that storyline, should be the answer to your question. There’ll be more new jobs and different jobs. There’ll be different jobs. If you get comfortable getting comfortable, in other words, get comfortable being uncomfortable, we have to learn about new things all the time, then you won’t have anything to worry about when it comes to employment.
Accuracy in Artificial Intelligence
George: Well, that is the second time I’ve heard chess analogies in the last month. The analogy that I heard was, “It’s good for your brain as you get older to play chess and keep that thing moving.” But thanks for that. It is fascinating. It is fascinating to see that it isn’t a replacement, if we embrace it and get comfortable with it, it could help us achieve things that we never before thought possible. If I’m hearing you correctly. Now, in part one, you spoke about the four pillars of AI. Let’s break down those and give them to the audience here, because we gotta understand this space more, and we got the expert on the call here. So Paul, would you be so kind to kind of help us understand those four pillars that you’ve spoken about?
Paul: Yeah, so maybe I’ll talk about these as the four pillars of responsibility. But I think I’ll start out with a former president, I think it was James Madison, but a US president. He once said that the circulation of confidence is better than the circulation of money. And in our podcast last week, we literally talked about, “AI is gonna decide if you live by die or try.” Among all kinds of other things. And the world has been maniacally focused on two things when it comes to AI. They’ve been maniacally focused on, “How accurate is my AI?” So with 98% accuracy, I can tell you that’s a CAD, with 90% accuracy, I can tell you that this particular mole on your arm is likely cancerous. That’s a melanoma. Okay, great. So we’ve been working on that, and we’ve been working on how fast can we build the AI. So you may have heard of companies like Nvidia, whose stock is Sword, they used to make these things called GPUs, they’re side processors that were used in gaming. But as it turns out, they’re incredibly useful for matrix multiplication, which is the math behind AI. We don’t need to get into that. But the bottom line is, we talked about how accurate is your AI, and how fast can I build it. But how confident are we in the AI? And so I recommend these four pillars because now we have to be accountable. We move from accuracy is no longer enough, to accountability. So your AI has to be fair, robust, explainable, and have lineage. And in those four things, if they excite you, I could share the kinds of problems or the kinds of things we have to think about on those four pillars. And now is the time for your business leaders. I am telling you, you will decide today, are you gonna be a good actor or a bad actor in AI. And we’ve seen bad actors, I’m not gonna mention their names, but they’re ubiquitous, big tech companies, really bad actors in AI, and I think that’s gonna change. And if it just doesn’t change by the way people spend their money, it’s going to change with regulation. So already the European Union has a point on AI, the GDPR has articles within it, which talks about the responsibility of AI. The United States has just released their Bill of Rights on AI, although it’s not law yet. Canada has done the stuff. Of course, we all know that governments kind of move at the pace of molasses, sometimes, around regulation. But I think companies wanna get in front and be good actors of AI, and fairness, robustness, explainability, and lineage, those are the four pillars you’re gonna invest in if you wanna be a good actor.
George: Well, and I’d like that you kicked it off with fairness, because to your point of bad actors.
Algorithim and Bias in Artifical Intelligence
George: It is, “Why did I get that notification just now?” I’m reading into what you’re saying. Did I get the notification because I really need to know that thing and the algorithm said that George really needed to know that at that moment, or did I get the notification because that platform has noticed that I haven’t been involved in the platform in a while, and it just fired me a little bit of dopamine to get me to go back into their technology? That I’m reading into what you’re saying, and I’m not naming any names. But is that truly fair? Was I given that notification because it really is going to help me, or was I given that notification to help shareholders of that platform?
Paul: Yeah. And I think the example I would say is, what kind of… So if you tuned into last week’s episode, and if you didn’t, I encourage you to do, I talked about how AI actually works, and I talked about how we feed data. So is the data we’re feeding it biased? And if it is, then what do you think the rules that the AI is gonna come up with are gonna be? They are gonna be incredibly biased. And as it turns out, I believe AI has a race, sex, culture problem. And I’ll give you some great examples on this effect. I talked many times last episode, this episode, how I can use visual inspection, so computer vision, to go and detect if a mole is cancerous. Now think about this business problem. Melanoma, cancerous, skin cancer, has been growing every single year for the last 30 years in the United States, and the same will be in Canada. And even if every American or Canadian wanted to see a dermatologist, we don’t have enough of them to do that. So this is where AI prediction, in this case, is the use case. We could go and look at a tumor or a mole, and predict if it’ll become tumor, if it’ll become cancerous. Now since we can’t all see dermatologists, in this world of selfies, where everybody’s kind of taking pictures all the time. Imagine taking AI to scan your pictures. Or if you have an un-identified mole on your arm, you scan that every week and the AI would be able to detect changes that are invisible to the naked eye, that your doctor would never figure out seeing them once a year, to attenuations or color of the mole, and start to predict if it’s becoming cancerous or not. We can also save a heck of a lot of money. And you know why? For every 10 lesions that we surgically remove for our biopsy, only one melanoma gets discovered. So every time we biopsy things, we get 90% of it, is a lot of unnecessary cutting and knifing, which is more anxiety for patients and more costs. Now, beautiful, right? AI can help us. What’s the accuracy of AI for that? It’s about 90%, that’s what the studies will tell you. 90%, amazing. 90% on white people. On black people, it drops to 72%. Why? Because the data used to train the algorithms were mostly European and Caucasian people. And so now you could see how this could completely affect patient outcomes. I’ll give you one more example. If I was to go and use AI to do translation from English to Finnish, and then back to English, and I put in, “She is a firefighter, and he loves the ballet.” That would come back to me as, “He is a firefighter, and likely that she loves the ballet.” I’m hungry, all right. But the ballet, why is that? Because if a million pages of which the AI reads to learn how to translate refer to firefighters as men, then the AI will believe that firefighters are men. And so whether I look at recruitment, at outcomes, credit risk, healthcare, you name it, we have got to get a handle on that data, because biased data will lead to unfairness in the workplace around live by die or try. And that’s the example I’ll give you on bias and fairness.
Storing Data in the Cloud
George: Well, Paul, it’s a phenomenal example, because you put it into some pretty clear terms and interpreted it for us. One thing I’ve noticed in talking to organizations, they were really thinking that we were gonna save a bunch of money in the cloud, and yet they’re disappointed to find out that they don’t. Why do you think that is? Why isn’t the cloud saving us money?
Paul: So this is interesting, ’cause that question came outta left field and it’s not AI. And I don’t know if you looked at my bio, but my last book was called Cloud Without Compromise. So I’m also a cloud person, and I love that you brought that up. Everyone believed that you could run to the cloud and save money. And there’s two big reasons why that’s not happening. One of them is, the cloud is a different way in which you architect applications. We call them distributed applications. And so people picked up their applications or their workloads, and move them to the cloud. Now that gave me some nice cost savings to start with. For example, test and disaster recovery, those are pretty easy things to do. I throw them up there. But the actual applications, we no longer build applications, we actually compose them. So we take a whole bunch of distributed pieces of logic and bring them together to make an application. So if you go to amazon.com, that’s actually about a thousand pieces of microservices which are these individual components, that are stitched together to create that application. When you move that to the cloud, it’s gotta be kind of cloud-native. And most people are running with the same monolithic code that they built on the service. That’s number one. So you’re not kind of building something to take advantage of something. You just put it up there, and so you’re only getting about a quarter of the savings that you thought you were gonna get. The other big thing is this. The reason why the cloud was so interesting, it was gonna save money, is, we said it was utility-like computing. If I turn the lights on in my office, they go on, I get billed. I turn it off, they go off, I don’t get billed. And so when you move to the cloud, you only get billed for what you use. Great. Except when you didn’t build these applications as cloud native and you moved them in the same old applications that were on premises, you ran them the same. And when they started to get all kinds of performance issues, what did people do? “We’ll just allocate more resources, and we’ll leave those resources on.” And that’s where people are way overpaying in the cloud. They don’t really understand the way that cloud should work. It’s a different type of application. And over-provisioning of resources is the number one thing that customers come to me and complain about, “Why so expensive when I’m supposed to save money?’ And there’s, “We gotta start and look at the applications you’re running, and how they’re built to run.” And that’s the reason.
George: Well, and the reason it came outta left field, is with 21 different books, and all of the content that you have online, there’s a lot of things that I wanted to make sure that we unpacked in our episodes, but you definitely delivered for the audience. Paul, I really appreciate that. And thanks for joining us for both shows, because it is something that comes up all the time. We wanted to understand more about AI. You gave us some great foundation there in our first episode last week, and now this week, covering off the age-old question, “Is AI gonna take over the world, take away our jobs?” I think you gave us some really good things to think about, where it really can enhance and improve our lives, and we have definitely… We’re a lot smarter having met you over the last two episodes here around that buzzword, AI. Paul’s Zikopoulos, vice president of IBM Technology Group Skills and Enablement. Thanks for joining us on the Conquer Local Podcast.
Paul: It’s been a pleasure.
George: So the question is, will artificial intelligence have a net increase in jobs, or will it eliminate jobs? And Paul believes, while there will be displacements, it actually will create more opportunity. For instance, AI and innovation have created new roles. He refers to it as new-collar jobs. With the advancement and regulations around AI, Paul reminds us to ask ourselves if we’re going to be a good actor or will we be bad actors when embracing this technology. For instance, if the data we’re feeding into the AI is biased, then the results we’ll get from AI will be biased. If you like this two-part series from Paul Zikopoulos, discussing artificial intelligence, let’s continue the conversation, and check out episode 501, Staying Human in the Age of Data with Rishad Tobaccowala, or episode 442, Mastering Predictive Sales: Data and Instinct, with Chris Bondarenko. Please subscribe and leave us a review wherever you listen to podcasts. And thanks for joining us this week on the Conquer Local Podcast. My name is George Leith. I’ll see you when I see you.