The AI Employment Revolution – Wharton Global Forum São Paulo

Hey, everyone. Thank you for coming. Have you guys seen this tweet before? Which I really thought encapsulated what we think of AI today. You know what the biggest problem with pushing all things AI is? Wrong direction. I want AI to do my laundry and dishes so I can do art and writing. Not for AI to do my art and writing so that I can do my laundry and dishes. Okay. So what's going on here? Are we doomed to do menial jobs? As AI becomes more advanced over time? What's going on? What is driving this amazing advance in AI, especially generative AI, the last couple of years? Well, generative AI in the parameters has grown exponentially. Why this matters, I'll show you. So back in 2018, GPT, the model that powered Chat GPT was only about 100 million in parameters. A couple of years later, before Chat GPT exploded, it was about 175 billions of parameters. Today, GPT Four and Gemini, those state-of-the-art models, have 1.76 trillion of parameters. So in a matter of six years we went from millions to billions to trillions. Okay. That matters because as a result of these many parameters, that we have never had before, many tasks are surpassing human abilities or reaching parities to human abilities. Okay, so the dotted line on top, the red dotted lines are human baseline. And these are a variety of tasks, benchmarks ranging from image recognition to natural language understanding, to competition math. So you can see not only are these benchmarks surpassing human beings, but look at how fast they're improving. This is competition math, Math Olympiad. Just a couple of years ago. It's a single digit inaccuracy. Now it's above 90%. Another one, MMLU, which is a very big benchmark, covering 57 subjects from STEM to humanities to social science and law. It covers factoids and all the way to complicated reasoning. Look at that. It has already reached human parity in a matter of few years. So the speed at which this is going is tremendous. So, of course, you look at this like, of course, am I going to be afraid of losing my job? Is Robocalypse coming? Well, popular press have written about this. Harvard Business Review, Foreign Affairs are talking about ÒWill robots take over our jobs?Ó Politicians, policymakers have summed, Well, maybe we should tax robotic humans? Bill Gates even thinks that is true. And this has been accelerated with generative AI. And just a couple of months ago, Bernie Sanders thinks that robots should be taxed just like humans. Okay? So are we doomed? Are we going to have a robot overlord telling us what to do? They're going to become so cognitively advanced that we will do menial work? Or is our future bleak? Before you despair? Let me show you what my research have done to show how AI and robots are actually affecting labor. So we studied the Canadian economy, 170,000 firms over 15 years. So we painstakingly track the robots, what they bought, how much you spend on them, and what the robots can do, what their capabilities are. And this is what we show. So the dotted red line is the beginning of adoption. Left-hand side is before adoption, and right-hand side after adoption. Instead of Robocalypse, firms adopting AI robots hired more people. So what's going on then? Right? I thought they were displacing people. It turns out that AI, even generative AI, still cannot do most of what we can do. And when you optimize some parts of our task, the demand for other things we do increases dramatically. As a result, we actually need more people. But if I show you, instead of employment count, but just managerial headcount, you see an opposite trend. Before robot adoption they're lots of managers. And after adoption there are lot fewer managers. Okay. It kind of plateaued after two years. Probably some kind of organizational change and then really plummeted afterward. Okay, so what's going on? Because the employment count belies a much more complex story about a skilled man distribution. If you divide into high, low, and middle skill work, you can see the demand for high skill work will continue to be strong. High skill will always be fine. But demand for low skill also went up dramatically. But middle skill work is being decimated. Okay. And large language models, generative AI, will accelerate this trend. Okay. So now we see an entire different skill distribution. That means you also need a new type of manager. So what's the big deal? I showed you we're going to have jobs. We have fewer managers, maybe not a bad thing. So what's the big deal? Well, you think about if you're a lower skilled worker, what can you do? I can work hard to become a middle skill worker. I can work hard to become a supervisor. But once I reach there, I don't have a job anymore. So the fundamental career ladder is broken. We don't work just for money. Of course, money is nice, but we need to have career aspirations. When the middle rung of the ladder is taken out, what happened to our career? How do you motivate workers? And that is a much bigger and much more important question that business leaders should think about. How do we rebuild this career ladder back again? We can ban AI. Many policy workers, I mean, policy makers have been proposing that. We should tax robots. But I just showed you that firms that adopt AI are more productive and hired more people. ItÕs the firm that did not adopt the AI and robots thatÕs getting killed by competition. As a result, they lay off workers. So these policies have the exact opposite effect. Instead, we should help. We should help firms rebuild that middle skill work. How do we make sure workers can effectively leverage AI? Think of it as a tool, collaborate with AI to repair robots, to work with robots, and find a way to optimize processes using robotics? Because the labor force is changing. That means a new type of management is needed. Long gone are days where the managers or supervisors dictating what you do, make sure you show up on time, tell you exactly what to do and how to do it. We need a more coaching type of manager. Because why? Because guess what? When you are using AI to replace processes, who knows them the best? They are the people who work on those processes directly, the line workers themselves. So they have deep tacit knowledge about the work. And they are the best to identify where AI could optimize things and where AI should never touch it. So youÕve got to leverage the skills they have to help you optimize AI. So what can you do as an individual then? Well, you know, AI is here to stay. It's not going to go away. I recommend that you just use AI in everything you do for a day or two. Okay. Think about what tasks I can optimize using AI. Where can you be more productive? Where not? And think about areas where you can be more effective, and think about how to use the free time you have to make you become more valuable in your role, in your organization. As leaders your job is a little more complicated. Complementing AI is the key. AI is not a plug and play where you say, okay, I want to optimize the processes I'm going to put AI into it. I'm done with it. That's not true. Often AI needs to change the organization. It has profound upstream / downstream effects. Even though you are just changing one little part of it. That means the entire new way of organizing work needs to be developed. In my research, I found that AI helped decentralize organization much more than centralize organization when it comes to innovation. As I mentioned, building that career ladder is very important. You may temporarily get rid of middle skill workers, but they are the most valuable for you as you optimize processes using AI. Their tacit knowledge matters. Okay? Ultimately, there will be disruptions and managing those disruptions, both short-term and long-term will be very important. Okay. So instead of thinking AI and robots as a competitor for your jobs, think of them as a tool, a great productivity tool. Think of them as your collaborators so they can help you do a better job at what you do and help you become more valuable in your role. So instead of thinking robots and AI to do our art for us, to write for us just like the tweets I have shown. We should think of them as a tool to help us write better poems, create more compelling art, advance science innovation, and ultimately create prosperity together. Thank you.