AI is here
Kellogg is training leaders on how to harness AI to solve business problems
When Professor Brian Uzzi describes Kellogg's philosophy for teaching artificial intelligence, he reaches for a distinctly low-tech analogy: fire.
"Fire changed everything about the human race," he says, from protecting ourselves from predators, to staying warm in cooler climates.
Those developments were, quite literally, life-changing. But, Uzzi explains, they didn't grow out of a deep understanding of the chemical properties of fire. They happened because people figured out powerful ways to use fire.
Which is how Kellogg is teaching students about AI. The role of a business school isn't to teach the underlying chemical properties, so to speak, of AI. That's the job of skilled computer scientists. Instead, business leaders need to know how to use this life-changing tool. This means having a comfort with AI and its applications so that you can use it to grow your organization and spark innovation, from decoding consumer data, to streamlining your supply chain, to revamping your hiring process.
"Students gain an understanding of the applications of AI," says Uzzi, a professor of management and organizations. "Then they're a long way toward being able use AI tools when they start companies of their own, being able to keep their companies ahead of rivals, and implementing their creative insights."
Kellogg is doing this through new MBA courses and executive education programs, collaborations with Northwestern Engineering, and increased opportunities for learning outside the classroom. The school is also investing in new faculty who, along with those already here, are using AI in their research to glean insights from massive data sets, as well as studying how AI will impact society.
The stakes for business leaders are high, Uzzi explains. As AI continues to permeate more and more fields, there are more ways to use AI to your organization's advantage, and more ways competitors may be using it to disrupt your industry.
"If you're a businessperson, you'd better understand what's happening and be able to talk confidently about it," says Uzzi, who is also the co-director of the Northwestern Institute on Complex Systems (NICO). "Otherwise, you're going to miss opportunities or be the victim of someone else's opportunity. So you've got to know what's going on."
AI's massive potential
The unlimited potential of AI is also its biggest challenge.
"There's hundreds, thousands, an infinite number of things you could do," says Adam Pah, a clinical assistant professor of management and organizations. "The question is, what could you do well?"
This is how Pah approaches his Human and Machine Intelligence course, one of several AI courses offered to MBA students. The others include Social Dynamics and Network Analytics, which focuses on using AI to understand patterns in big data, and Visualization for Persuasion, which tackles how to interpret and present information gleaned through AI in compelling and convincing ways.
Pah, who is also the associate director of NICO, focuses on understanding where machine intelligence excels today, and where it is headed in the future. He also helps students hone their skills as leaders who will be confident choosing which operations should be automated and which won't benefit from AI.
"All too often, people think about artificial intelligence as this really neat tool. But if you make something no one's asking for, you haven't really done much of anything."
The course ends with a case study on how a specific company could use AI to improve its operation. "It forces the students to take all of their learning and reduce it down to implementable practice," Pah says.
He has done this by teaching an award-winning case he wrote about Honeywell Aerospace's decision on whether to use AI to manage aircraft maintenance, and by bringing in executives to discuss their companies' challenges. For example, last quarter, Pah brought in Kellogg alums from NetraDyne, a startup that makes cameras that record and analyze commercial truck drivers and the routes they're driving so that drivers can receive real-time feedback on their safety performance.
The students recommended a way for NetraDyne to repurpose its camera data of the drivers' routes, which shows the roads' condition and the volume of traffic on them. They suggested using this data to help cities predict how long a specific road repair will last.
Such exercises helps students understand that AI is not simply a shiny new thing to chase around.
"All too often, people think about artificial intelligence as this really neat tool," Pah says. "But if you make something no one's asking for, you haven't really done much of anything."
Teaching executives
Kellogg executive education has seen a spike in demand from business leaders who want to tackle many of these same issues.
"They're trying to figure out how to deploy AI in their organizations," says Paul Christensen, the associate dean of executive education. "They want to build those capabilities in their organizations, and they worry a lot about the potential for disruption that it has in their industries."
In the past year, executive education has put together custom online programs for two large global companies on data analytics and artificial intelligence, which thousands of their employees have taken. Another roughly half dozen companies have participated in custom in-person programs.
"Our approach to teaching AI is to be completely driven by the problems that you're trying to solve," says Florian Zettelmeyer, a marketing professor and faculty director of the Program on Data Analytics at Kellogg.
Those may be problems in customer marketing, or automation, or computer interfaces, he says. In both custom and open-enrollment programs, Kellogg helps business leaders come up with AI-driven solutions and ensures they have the under-standing to make those solutions work.
Often, custom clients come to Kellogg to help spread a comfort with AI around their organization, Christensen says. For example, they may have a large and skilled data science team that's ready to roll out a predictive algorithm, but other leaders in the company are skeptical. So one custom program helped establish an organization-wide understanding of machine learning so that everyone could work more confidently with the data science team and use AI to its full potential.
Another custom program came from a company that asked for a second, more advanced program after putting hundreds of managing directors through an initial Kellogg training. The new "201" version has students learning interactively to do data analysis and run experiments in real time, Zettelmeyer explains, with realistic assignment prompts such as "OK, you've done your data analysis, and now your bosses ask how you can prove to the client that this is valuable."
One of the largest open enrollment programs in executive education is Leading with Advanced Analytics and AI, taught by Zettelmeyer and Eric Anderson, also a professor of marketing. The program provides senior leaders with a working knowledge of data science so that they can identify business problems that analytics and AI can solve.
This year, executive education is also adding two new open enrollment offerings.
Digital Marketing Strategies: Data, Automation, AI and Analytics is taught by Mohan Sawhney, a clinical professor of marketing and director of the Center for Research in Technology & Innovation. The online program is for business leaders, managers and marketing professionals who want to learn more about how to apply these tools to marketing and product design.
The other program, Leveraging Artificial Intelligence for Innovation and Operational Efficiency, will launch in the spring. It's being taught by Uzzi, Pah and a number of other faculty, including David Ferrucci, the researcher who founded and led IBM's Watson team and is now an adjunct professor of Entrepreneurship and Innovation. This program will focus on how to use AI to drive growth and spur innovation, including how to build human-machine teams.
Many executive education programs involve partnerships with computer science professors at Northwestern's McCormick School of Engineering, ensuring that executives have a baseline understanding of the more technical issues.
For example, Zettelmeyer says many companies are experimenting with natural language processing for their customer service needs, such as setting up chat bots. While executive education students do not need to learn how to create the code for these bots, the McCormick professors can provide an understanding of what the technology can deliver and where it is headed in the future.
"These have enormous implications for companies because they are potentially huge cost drivers," Zettelmeyer says.
The student experience
Collaborations with McCormick have also played a large role in giving Kellogg MBA students more exposure to artificial intelligence outside the classroom.
Last year, Pah joined forces with Kris Hammond, a professor of computer science and the director of McCormick's new Master of Science in Artificial Intelligence, to bring students together for that program's "Industry Nights." One night a week, MBA and computer science students gather to hear from professionals who talk about their companies. Then, the two schools' students work together to come up with a practical AI-based proposal for that company.
Many of the Kellogg students who attend these events are part of a new Artificial Intelligence club, which was created by Kenn So '19, an MBA student who graduated this spring.
So also took advantage of the ability to take courses at McCormick, which was one of the reasons he chose to come to Kellogg. He took classes on coding and machine learning, and another called Algorithms and Society that gets students thinking about how these technologies affect our lives. It's a topic that he's fascinated by and has continued thinking about after graduation.
And he knows he's not alone in that interest. During his two years at Kellogg, he saw more and more students join the AI club, eager to spend time outside of the classroom thinking about the technology's impact on the business world.
"Everyone's interest is piqued by artificial intelligence because it's everywhere," he says. "Everyone wants to know more."
Faculty focus on AI
That "everyone" includes faculty.
In their research, Kellogg professors are approaching AI in two ways: some are using powerful AI tools to answer novel research questions, while others are putting AI itself under the microscope, studying how it is impacting society.
Kellogg's new interdisciplinary Center for Science of Science and Innovation falls in this first bucket. Faculty like Uzzi and Dashun Wang, an associate professor of management and organizations, use machine learning algorithms that can crunch data and tease out patterns far faster and better than humans can on our own.
For example, Wang and coauthors created a model that can predict how popular individual items or ideas will become. They tested their model on 100 years' worth of physics research, focusing on the citations that each paper garnered. (Citations are the academic paper's version of a popularity scale.) They found that by training their AI-driven model on 10 years' worth of papers, it would go on to predict the next 20 years of citations more accurately than other models. And the model is flexible enough to be used in other domains where popularity is important, from marketing a product to getting a hashtag to trend.
This sort of analysis would not be possible without AI tools.
"We are living in an era where big data permeates every corner of society," Wang says. "We're using data science and artificial intelligence to make sense of this data."
In the marketing department, recent research from associate professor Jennifer Cutler can help companies better understand how their customers feel about their brand. She and a co-author developed an algorithm that analyzes Twitter activity in real time. It uses information about who follows whom to calculate a score that can tell companies how much people associate their brand with a certain attribute, such as being environmentally friendly, or luxurious.
Other professors have examined the economic and cultural changes that AI will surely usher in.
For example, Hyejin Youn, an assistant professor of management and organizations, has researched how automation will impact different US metropolitan areas. After all, automation will undoubtedly have a bigger impact on some jobs—say, cashiers and accountants—than on others including, scientists and software designers, for example. She and coauthors used data on the likelihood that different jobs would become automated to predict how many workers would be displaced in 380 metropolitan areas across the country.
She found that, overall, smaller cities are going to be harder hit by automation. And this automation will likely lead to greater income inequality because the jobs that will be spared from automation are either very specialized and high income, such as a lawyer, or low income and therefore not worth investing in automating, such as a janitor.
She hopes that policy makers use research like hers to help soften the blow of the coming wave of automation.
"If I'm a policy maker, then I have to think about how to reshape the industry in my area," she says. "And I need to think about it in a detailed, nuanced way. Not just: automation comes in and hits the whole US"
One tool policy makers could consider is a robot tax, which finance professor Sergio Rebelo has studied.
Bill Gates proposed this tax to replace the tax revenue lost when jobs are automated. Rebelo and coauthors found that automation increases economic output and tax revenue, so there is no need to tax robots to replace lost revenue. There is, however, a different reason to tax robot use, which is to improve the distribution of income. Such a tax only makes sense before routine tasks are fully automated. Once full automation occurs, a robot tax hurts economic efficiency without improving income distribution.
Adam Waytz, an associate professor of management and organizations, is interested in how humans learn to trust machines, as well as how we understand a machine's moral responsibility for wrongdoing. In one study, he found that people were more likely to trust an autonomous car when it had human-like features, such as a name and human voice. And people were less stressed during a minor accident caused by the anthropomorphized car than they were in an autonomous car without the human-like features.
"We are living in an era where big data permeates every corner of society. We're using data science and artificial intelligence to make sense of this data."
Bringing in new faculty
In addition to these and other faculty studying AI, Kellogg is working to expand its faculty bench. To that end, the school this year hired three new faculty whose research either investigates artificial intelligence or uses AI tools.
Hatim Rahman joined the management and organizations department this fall after receiving his PhD from Stanford. His research explores how algorithms and other types of artificial intelligence are shaping the future of work.
Sebastien Martin and Artem Timoshenko are both joining the faculty after completing PhDs at MIT.
Martin, who is joining the operations department, has used artificial intelligence algorithms to improve school bus efficiency in Boston, saving the city $5 million a year. Timoshenko, who is joining the marketing department, uses artificial intelligence tools in his research to help retailers analyze customer data and user-generated content.
Driven both by an interest among business leaders as well as fellow faculty, Kellogg will continue to recruit researchers in this area, says Michael Fishman, the senior associate dean for faculty and research. And the school plans to increase its partnerships across the university when it comes to researching and teaching AI.
"These approaches to understanding data are becoming more and more important in the business world," Fishman says. "It's not a fad. It's not going to disappear."