We asked three Booth experts: alumnus Justin Adams, ’10, professor Nicholas Polson, and PhD candidate Diag Davenport.
- By May 09, 2019
The question is, how do you take the horse-and-buggy driver, or the accountant today, and give them new skills?
So on a larger scale, it will be good, though it may cause some people to lose jobs in the shorter run. If you look at big game changers from a societal standpoint—the Industrial Revolution, automobiles, the internet—there’ve always been more jobs created than lost. Since 1990 over a billion people have been lifted out of poverty globally, largely driven by capitalism and technology. The horse-and-buggy driver had to find a new profession when cars came along. At the micro level, there will be disruption; to deny this is naive. From a larger, societal perspective, I’m optimistic AI will follow the same trend.
So the question is, how do you take the horse-and-buggy driver, or the accountant today, and give them new skills? I think a lot of jobs will require more soft skills. As with anything, unfamiliarity breeds fear, and maybe contempt. For people who are worried, sign up for online classes in AI, just to know what the heck it is and get familiar. AI is a catchall phrase that’s exciting and scary, but the concept has thousands of subcomponents. At a high level, be able to understand the opportunities, and from there, think it through. Assume AI is going to augment or replace parts of your job. Identify what those are beforehand and be proactive. This will add value for the long run. Don’t be afraid that it’s going to replace your role. Be prepared.
Nicholas Polson is the Robert Law Jr. Professor of Econometrics and Statistics and coauthor of the book AIQ: How People and Machines Are Smarter Together.
Lots of things are changing, and have already changed. Almost every company is using AI in some form these days. Among the earliest to embrace AI was Netflix, one of the original recommender sites. The AI algorithm there collects big datasets of people’s likes and dislikes in movies, and does pattern matching. This is all it is, a system of prediction. Now Facebook and Spotify and lots of other companies use pattern algorithms to come up with better recommendations.
Then there’s AI like the robots in Amazon warehouses. And one of China’s largest ports, Qingdao, is totally automated. The port in Rotterdam, Netherlands, is mostly automated. These places would have had thousands or tens of thousands of people working. Of course, the people running them think it’s great.
There’s a video circulating on YouTube—another great prediction system, by the way—of a Volvo truck stopping inside of 30 yards to avoid hitting a little girl in the road. These trucks have an automatic emergency-braking system, which is more sensitive and faster than any human driver. Everyone thinks automatic cars are dangerous, but they’ve got an incredible number of sensors. So while that’s a good thing, they’ll probably replace truck drivers one day. Imagine all the jobs lost. Once driverless cars are more common, car accidents will be cut down enormously, which will also mean that you don’t need as many emergency-room staff. ERs will be empty. They’ve already rolled out driverless cars in Arizona, so this is happening.
AI is better than the top radiologists, yet radiologists still exist. Maybe this is an example of people and machines working together, at least for now.
Certainly health care is a field where AI has a lot of applications. It used to be that radiologists studied for many years and accrued huge educational debt, but had a good job after. Now AI is better than the top radiologists, yet radiologists still exist. Maybe this is an example of people and machines working together, at least for now.
I guess the polite answer is that it’s both good and bad. On the positive side, it may free people up to do more of what they want. It’s freed me up, but I also write more papers now, and I’m more addicted to screens. So, in a way, it’s more efficient, but you get more caught up in stuff. Sometimes I prefer the old, less efficient ways. Now you’re on call 24-7. I think people are way more productive. But people are probably a lot less happy.
That’s the negative side—the Instagrams, the Facebooks, the Spotifys. The already existing AI products offer an incredible amount of leisure. But I call it infinite content, where you’re connected 24-7. A German neuroscientist coined the term digital dementia for the cognitive effects that occur in people who play video games all the time.
As far as preparing for these changes, that’s hard to say. There are people such as [Tesla CEO] Elon Musk who believe AI will take over everything, and people who believe it won’t amount to anything beyond computer chess. A lot of people are also encouraging their kids to learn to code. I’ve read that something like 45 million people are using the Codecademy website. But that’s a lot of people coding—and how many jobs are there? Machines are also coding. I’d say learn about what it can do. A lot of people misunderstand what AI is.
In the 1940s the British mathematician and computer scientist Alan Turing made a prescient speech about how machine learning would evolve and what effects it could have on jobs, both positive and negative. We’ve started to see how AI has changed jobs, and life in general—but we don’t know quite how it will play out in the future. So these questions have been around for many years, and will be around for a long time to come.
Diag Davenport is a PhD candidate in behavioral science.
I think it’s helpful to make the distinction between machine learning, or ML, and AI. Machine learning is a process where you take data and let some system figure out how it relates to the outcome—in other words, an algorithm. AI usually involves integrating ML into some hardware to do things—Apple’s Siri is a good example. So when it comes to playing a song I like, ML would make the predictions, but the task of opening the program to play it would be AI’s.
My expertise is ML, which is being introduced all over the place. Some people may be inspired by the idea but may not have a grasp of how best to use it. Others may have a strong or unwarranted aversion to algorithms. But it’s really the state of the world right now. There’s a lot of opportunities to automate, and it doesn’t have to be scary. I’m very optimistic, and think it’s a good thing that ML is out there in the ecosystem. And once it’s less novel and less a fad, it will be integrated more seamlessly.
Naturally I see ML replacing the repeatable tasks, the way a calculator replaced doing math by hand, which was a benefit. For example, one of an accountant’s tasks is to add up entries—but another, maybe more important one is to interact with the client and understand what to do next. Accounting and related fields can be bolstered by this very good prediction system that’s ML. Humans are good at interacting, but they’re not as good at calculating. ML is good at finding patterns in data, but it can’t interact with humans as well.
Managers should think of ML more as a calculator and less as a competitor.
I’m fairly certain there are tasks ML will never replace, including face-to-face interaction. Any industry that’s characterized by clear, repeatable tasks, with little human interaction, such as manufacturing, will be affected. Fields such as teaching won’t be. You have to have kids in a classroom interacting with a human teacher. Kids won’t learn at nearly as fast a rate if they’re interacting only with ML.
I would say that the best way to deal with these changes is to embrace them as best we can. In some circles there’s a lot of fear, which is probably not necessary. Managers should think of ML more as a calculator and less as a competitor. You can use ML as a decision aid, and it can get things you don’t want to do off your plate. Using ML as a calculator does mean you have to learn how to use a calculator, however—that’s the level of fluency people really need to have. But you don’t need to build a calculator.
Some jobs will probably be at stake—that’s unfortunately how it usually works when new technology comes about. Previously “go back to school” might have been the only advice. To the extent that you have the luxury to, think about the special thing that you add to the economy, or would like to. What do you want to specialize in? Start by looking at content on YouTube, LinkedIn, or MOOCs [massive open online courses].
To the extent that you can be patient, take advantage of the resources that are out there, and position yourself to gain from the increased productivity from ML that is on the horizon.
We'd love to hear your Booth memories, stories, connections...everything.