Intelligent Automation: Here’s How it Will Impact Jobs and Cities in the Age of AI

Morgan Frank, Researcher, MIT Media Lab and Co-Author"
Morgan Frank, Researcher, MIT Media Lab and Co-Author “Small Cities Face Greater Impact from Automation”

MIT Researcher Morgan Frank takes us to school on the astonishing evolution of intelligent automation and its impact on people and jobs in cities across the country.

Frank co-authored a breakthrough study called “Small Cities Face Greater Impact From Automation“, which empirically connects the impact of automation and urbanization on employment.

Here’s what the study revealed: Smaller cities—with less than 100,000 people— will experience the most disruption from artificial intelligence and other emerging technologies.

Why? because small cities tend to have a disproportionate amount of routine clerical work, such as cashier and food service jobs.

All of which are prone to compete with automation.

In this thought-provoking interview, Frank drops some serious knowledge on:

  • Why workers in larger cities will experience less disruption from emerging technology
  • Jobs least likely to be replaced by technology
  • The difference between competing technology and augmenting technology—and how each impacts the workforce stack
  • The rising challenge of occupational polarization.
  • Why intelligent automation will force your organization to rethink how work gets done

Hope you enjoy the conversation.

Appian: Good morning. And welcome to Digital Trailblazers. To get things started, tell our readers about your expertise and the research you’re doing at the MIT Media Lab.

Frank: Sure, my background is in computer science, statistics and applied mathematics. But my focus is in computational social science, which is using data and mathematics to understand human behavior and society.

In particular, I’ve been investigating the impact of automation on labor in the U.S., and especially how automation impacts cities.

“At the MIT Media Lab, we’re focused on drilling down into the relationship between occupations and skills and technology. Then, we try to identify which occupations will benefit from the complementary nature of new technologies, and which (jobs) are at risk of competing with them.”

The future workforce strategies, industries overall graph

Appian: In some circles, there’s a lot of fear about robotics and automation and how they will disrupt the future of work.

Critics are talking about the coming disruption of machine learning, algorithms, chatbots, voice recognition and intelligent automation. What do you make of that pessimistic view?

Frank: Based on our research at MIT, I think that the automation trend is not so alarming—not in terms of current technologies and technologies in the pipeline today.

But there are certain jobs and certain industries that should be concerned about automation.

Appian: Can you give us some examples?

Frank: Yes. For example, truck drivers are in direct competition with autonomous vehicles, which doesn’t appear to be a complementary technology for truck drivers.

“But what’s making the current automation conversation different from previous conversations is artificial intelligence—trends like machine learning, and digital assistants. And what’s different is that these technologies complement highly-skilled people.”

Appian: In what ways?

Frank: Machine learning, for example, is making computer programmers more efficient at understanding data.

It’s not competing with them. Computer programmers are not losing work because of machine learning.

(In contrast), truck drivers are in competition with self-driving cars.

So, there’s a juxtaposition between computer programmers and truck drivers.

Appian: So, let’s step back and look at the big picture. Tell us about your research on the impact of automation on urban areas versus rural areas. You’ve said that some small cities will benefit from AI and automation, because they are near major institutions that employ skilled workers—such as military bases, major universities and corporate research labs.

So, where do you see automation having the biggest benefit geographically?


“With AI-related and cognitive technologies, we’re seeing a lot more benefit in larger east coast cities in terms of augmenting the skills of highly-skilled labor in industries like tech and financial services, which have lots of high-skilled, knowledge workers.”

We’re also seeing the same thing in Silicon Valley, in Los Angeles, Portland and Seattle.

Appian: So, net-net, do you expect automation to create more jobs than it displaces?

Frank: That’s a super tough question to answer. For anyone who is seriously studying these trends, it’s hard to make a confident prediction.

So, one way to think about this is that job creation definitely occurs with new technology.

But it’s just so hard to predict.

We’ve got a quote from former IBM president Thomas Watson [1943] that there was a world market for maybe five computers.

But today, everyone has a computer in their pocket or backpack.

If I could predict the new employment opportunities that will come from new technology, I wouldn’t be writing research papers [laughter], I’d be starting a company.

Appian: You’ve also said that one of the biggest impacts of automation is that it’s redefining work. What did you mean by that?

Frank: The other side of this, is that technology often changes the skill requirements for specific jobs.

And this can change fluidly over time, to reflect demand for individual tasks and skills. So, think about the robot arm on an assembly line.

It’s designed to do something very specific.

And if you’re talking about machine learning, there are machine learning algorithms designed to solve a specific class of problems. So, each algorithm is somewhat narrow in scope.

“The thing is that automation is happening at the level of specific tasks and skills. And the changing demand for workers that can perform these tasks and skills tends to bubble up as disruption of employment and wages in the labor market.”

Appian: Can you give us some real-life examples of that kind of disruption?

Frank: There’s an economist at Boston University, named James Bessen, who made an important observation about bank tellers, and automated telling machines (ATMs).

You might expect that with the ubiquity of ATMs, national employment for bank tellers would go down as ATM usage rises.

But the opposite is true. Bessen showed that overall employment for bank tellers increased proportional to the increased use of ATMs.

That surprised many people. But there are two reasons why this happened. First, the efficiency of ATMs made it cheaper for Banks to open branches.

Fewer tellers were employed per branch. But there was an overall increase in nationwide employment.

The other factor to consider is that the tasks traditionally associated with bank tellers fundamentally changed.

It used to be that they did mostly clerical, routine, work. Count money. Give you change back.

But ATM’s freed them up to act more like customer service representatives and salespeople for bank products.

Appian: So, tellers transitioned from clerical work, to doing higher-level work that adds more value.

Frank: Yes. And that kind of work requires more social skills. This is the kind of occupational redefinition that I mentioned earlier.

And we’re trying to understand this and model it even better.

Appian: So, it sounds like the impact of automation less about job displacement than about rethinking how work gets done.

Frank: Exactly.

“In addition to job creation from new technology, there is also a lot of redefinition going on. And that’s the bulk of the impact of automation (on the workforce).”

Labor Versus Smart Tech

Appian: That said, what can traditional companies do in terms of how to prepare their workforce for intelligent automation?

What steps can companies take to prepare for the smart tech that’s coming down the pike?

Frank: The thing is to understand who in your labor force is going to be in competition with emerging technologies.

Let’s say you operate several factories. And a new robotic arm will allow you to reduce jobs related to a specific assembly line task.

The question is, what will you do with the workers that you will no longer need?

Appian: The traditional approach would be to lay them off.

Frank: Yes. Or you can retrain them to work at one of your other locations to perform tasks that you still need human labor to perform.

But this requires a deep understanding of your workers’ skills—what’s going to be replaced by technology?

What skills will still be in demand? And how will you match the skills of your workers with expected demand?

Appian: So, what’s the secret to doing that? What’s the best approach?

Frank: Right now, there’s no road map. This is something that the research community is trying to identify.

Based on aggregated trends, I can make some general recommendations.

For example, if you can teach your workers to work with computers, and perform cognitive and social tasks, then they are more likely to be augmented by technologies that we now see in the pipeline.

Over longer time frames, social skills are a special topic that the research community is interested in.

It seems like social skills really don’t have any competition in technology.

“It’s harder for technology to replace human interaction in many important ways. For example, I don’t think customer service will go total automation anytime soon.”

And also, the way deals are struck seems to rely more on networking and social capital. And this isn’t something that’s going to be automated away.

Appian: What were some of the surprising things you learned from your research on automation and its impact on cities?

Frank: The cities that are surprisingly resilient to the rise of automation are small cities that house universities or a critical mass of government offices.

Appian: Any specific cities come to mind?

Frank:  Yes, Burlington, Vermont and Boulder, Colorado. These are relatively small major cities.

But they are outliers on many important economic indicators—like GDP per capita, and health and well-being.

Universities are driving local economies that can support demand for high skill workers—workers who can use cutting-edge technology rather than compete with it.

In Boulder, we’re seeing a “Latte” town that’s growing into a larger city. I say that because we’re seeing an increased presence of the tech industry in Colorado.

Google is opening a big office there. Facebook and Twitter also have a presence there now. And I think this will change things for Boulder.

The Best Technology Works like Magic

Appian: From an emerging tech standpoint, we’re still in the early stages of digital transformation.

But what are your expectations in terms of what the relationship between people and technology will look like in the future?


“I think it’s going to become increasingly hard to recognize that you’re working with technology. At the Media Lab, for example, we think that working with really good technology is like working with magic. That’s the type of interface we want. You’ve really nailed it when you can do that.”

Appian: Modern smartphones are kind of magical like that.

Frank: Yes. You can pull this little cube out of your pocket, and out of thin air, you find virtually any information you want, you can just pull it out of thin air. And there’s something sort of magical about that.

AI Goes Beyond Information Retrieval

Appian: What are your expectations for artificial intelligence (AI)?

Frank: So, I think we’re going to see more and more AI and algorithms that are doing more than just information retrieval.

They will become a bigger presence in our daily lives. And you might not notice it so much, because of the gradual change.

In the Media Lab, psychologists are working with children to study how they interact with things like Amazon Echo and other digital assistants.

We’re looking at what these interactions do to their social skills and their expectations for interacting with other “dumber” technologies.

Appian: And what are you learning from these observations?

Frank: So, what we’re seeing is that children are willing to interact with digital assistants. But they’re a little confused about whether the technology is a person.

“As adults, we’re going to see more instances where data about us is being used without us knowing it, to drive conveniences that we expect. Amazon Echo is a good example of that. It can play the radio shows that you like. It can tell you about the traffic or the weather. And it can do these things automatically.”

Regulators, Behind the Digital Curve

Appian: What do you make of critics who fear AI and worry about the misuse of it?

Frank: There will be a down side. I can’t provide you with the science on it. But I do believe fake news will get worse. And as a society, we need to solve this problem.

Usually, scientists develop new technology and leave it to the regulators to regulate how it’s used. And that’s healthy and good.

But I don’t think that regulators are keeping up with emerging technologies. And they need to pick up the slack.

Appian: Looking at 2018 and beyond, give me your top three expectations in terms of the digital trends that are on your radar.

Frank: I think that occupational polarization is going to increase in the next few years. Part of this will be due to the distinction between competing technology and augmenting technology.

Also, I think many economists will agree that the impact of technology is increasing at an increasing rate.

So, from an educational policy perspective, this is a hard problem, because public policy is even more sluggish than science.

“In a world of constant accelerations, it can be difficult to make decisions about your four-year degree, and how to prepare for a workforce that’s four years out.”

So, I think improving educational policy in the face of fast changing technology is going to be huge.


(This blog was originally published February 26, 2018 and updated December 5, 2018)