Inhuman Touch

The Inhuman Touch: Educators Teach the Nuances of Artificial Intelligence

By Sharon Shinn

December 20, 2016

Artificial intelligence is coming to the business school classroom. What can students learn from the machines?

BIG DATA. ARTIFICIAL INTELLIGENCE. MACHINE LEARNING. Cognitive computing. They’re terms that are tossed around in business schools today almost as often as words like finance and marketing—and if they’re not quite foundational topics in business schools, they soon will be.

Big data, of course, refers to any massive amount of information that can be gathered, sorted, and analyzed to help managers make smarter decisions. The other terms refer to ways that computers are learning from and adapting to human behavior to make data even more useful.

According to James Pang, a visiting associate professor at the National University of Singapore’s School of Computing and Business School, there are four fundamental technologies in big data analytics: artificial intelligence; machine learning; information management; and massively parallel processing, in which multiple processors work on different parts of a program while communicating through an interface. He adds, “Because of recent developments in these fundamental technologies, we are able to deal with big data through the four V’s—that is, volume, velocity, variety, and veracity. This allows us to derive business insights from massive data sets.”

Until recently, many analytics programs were housed in colleges of engineering and computing, but more and more business schools are either collaborating with those colleges or creating analytics programs of their own. That’s because it’s important not just to analyze the data, but to use it to drive business decisions.

“Data specialists need to be great at managing data, but they also need to understand business issues,” says Cheri Speier-Pero, professor of information systems at Michigan State University’s Broad College of Business in Lansing. “Ultimately, their decisions will affect a company’s marketing strategies and financial outcomes, so they need to be effective in creating and making sense of statistical models.”

For business schools, two questions arise, says Speier-Pero. “How do we prepare our students? And how do we create interesting and cutting-edge knowledge by applying these techniques to our research?”

NUS and MSU are among a growing number of schools that have taken those questions seriously. They’ve both launched partnerships with IBM and other software companies to give their students a head start on using the computing tools that will be so essential in the workplace. (See quick descriptions of their programs in "Two Approaches to Teaching Analytics" in the sidebar at the end of this article.) And running those programs has given these two educators insights into what the marketplace needs—and what the future holds.

To prepare students to work effectively in the big-data future, one of the first keys is to familiarize them with cutting-edge tools. That’s why schools find it essential to partner with big players like IBM and SAS, which can provide access to the most current technology and give instruction on how to use it. Most schools also follow a multidisciplinary approach that often combines business, computer science, and engineering. But Pang and Speier-Pero both emphasize that schools must do more than teach students about technology.

Pang believes the first step is to train students to embrace the future with open minds; the second step is to encourage their ability to learn quickly, because the world is changing at unprecedented speed. “The third step is to make sure our students understand the fundamentals of business,” he adds. Thus, students must learn how to identify and understand “pain points” in business, he says, because these represent chances for innovation. Students also must master essential soft skills such as leadership and communication.

Speier-Pero believes it will be just as important for students to know how to learn on their own. “While tech companies and employers have gotten better at providing training on an as-needed basis, they don’t hand-hold students through the process,” she says. “We tell students they will never be given sufficient training in a given software, so it will always be partly up to them to figure out the software. At the university level, we need to instill in our students this approach to learning so they will be more successful in the workplace.”

Students need this kind of preparation, say Pang and Speier-Pero, because companies already are employing AI techniques to learn human behavior in fields such as banking and finance. To help her students understand how AI is working in the real world, Speier-Pero begins with a topic most of them understand: credit card fraud. One way to avoid fraud is to put chips in the cards, she says—but another is to learn a specific cardholder’s shopping behavior and identify aberrations, and that’s something machines can help humans do.

“I use my mother as an example,” she says. “Every year, she goes on a shopping binge that is different from her traditional way of shopping. At first when those binges would occur, the company would call the house to see if my mom was the one using the card. Now the company has realized that behavior is normal for my mother, even if it’s infrequent, and built it into her profile, so the binges no longer trigger warnings of fraud. To me, that’s a cool example of artificial intelligence because the machine is looking uniquely at a single shopper and learning her behavior.”

But many other industries potentially can be transformed by artificial intelligence. While Pang warns that it’s difficult to make predictions in the rapidly changing world of technology, he looks for AI to have a great impact in four key areas in the next three to five years:

Personal digital assistants. Smartphone assistants such as Apple’s Siri and the Google NOW app will learn more and more about our behaviors. “Assistants will provide us with more personalized services, such as recommending movies and products, and will arrange our daily schedules,” says Pang.

“Autonomous transportation, or driverless cars, soon will be commonplace,” says Pang. “For most people, they will be the first experience with physically embodied AI systems, and they will strongly influence the public’s perception of AI. Many companies—including Google, Uber, Tesla, and Baidu—already are making significant progress in this area. The world’s first driverless taxi already is operating in Singapore, and Uber also has started its first driverless taxi service in Pittsburgh.”

Healthcare. “AI has already started to make an impact in this industry,” says Pang. “For example, the IBM Watson Oncology Advisor now provides oncology treatment advice to doctors at several hospitals in North America and Asia. The journey of AI in healthcare will not be fast or smooth, but healthcare will be one of the most impactful AI applications in our daily lives.”

Home/service robots.
“These already have entered our lives as intelligent vacuum cleaners and servers at restaurants, but we’re going to see more of these,” Pang predicts.
He notes that it’s hard to forecast ten to 15 years out, but in that timeframe he expects AI also to make an impact in areas such as public safety and security, employment and the workplace, education, and entertainment. All of these are fields that will be seeking business graduates trained in the power of analytics.

Some observers shudder when they see these lists of industries that are being taken over by tech, fearing that machines will displace even more people from their jobs, but both Pang and Speier-Pero have more optimistic views of the digital future.

“I don’t think AI machines are threats to humankind. I think they will have profound positive impacts on our society and economy,” says Pang.

Speier-Pero agrees. “I really believe machine learning is making people more valuable,” she says. “In fact, their higher-level skills might be more critical than ever. The machine environment can perform the analysis faster than the human can, but the human can understand nuances better.”

The challenge remains educating students to work with machines so they can leverage both artificial and human intelligence. For most people who are intrigued by AI, says Speier-Pero, the real obstacle is that technology is being applied to such complex problems.

“These problems are very costly from an organizational standpoint, so they’re worth looking at, but there’s tremendous learning required,” she says. “To me, there are interesting ways that machines can learn to better understand human behavior that will allow us to improve financial results for companies or experiences for customers—ways that don’t require a high level of sophistication. My guess is that, once some of those approaches have proved successful for companies, they’ll start thinking about more sophisticated uses for AI.”

And that’s when business graduates need to be ready to combine their business acumen with their technical savvy to lead their companies to success.


Two Approaches to Teaching Analytics

Many schools have recently launched business analytics programs. Here’s a look at two.

■ At the National University of Singapore, the master of science in business analytics is a collaboration between the NUS Business School and the School of Computing, and it includes a module built around Watson Cognitive Computing. IBM experts come to the NUS campus to deliver technology seminars and hands-on workshops to explore how to apply cognitive computing in real-world business cases.

In addition, students from the business and computing schools form teams to build creative Watson applications for specific industries. In 2015, the school launched an event called Watson Innovation Challenges, receiving 34 student proposals covering 14 sectors, including healthcare, retail, and human resources. “Some of the interesting applications students built included a personalized treatment advisor for rheumatoid arthritis, an online shopping advisor for consumer electronics devices, and a digital recruitment expert for HR professionals to use in hiring,” says James Pang.

By the time students leave the class, Pang hopes they’ve gained three types of knowledge: an excellent grounding in fundamental and state-of-the-art technologies in cognitive computing; an understanding of how to use technology to build solutions to “the pain points” in real businesses; and the ability to work with industry professionals to deliver a successful project.

Pang notes that it’s essential for the class to keep evolving because the technology in the area of cognitive computing is changing rapidly. “In the beginning, students developed applications in a dedicated Watson Cloud for NUS, but now Watson capability is available from IBM’s new BlueMix platform, which students leverage to develop Watson applications,” he says. Bluemix allows organizations to tap into IBM and third-party services for free on a public cloud, or to run private applications behind a firewall. (Learn more at

Michigan State University’s master of science in business analytics is a three-semester collaboration among the Broad College of Business, the College of Engineering, and the College of Natural Science, which houses the statistics department. The school partners with companies such as IBM and SAS to give students experience with statistical platforms and programming languages.

One of the critical pieces to MSU’s program is a co-curricular project that students begin in January before they enter the classroom. All students are assigned to teams; they all work with data that comes from the same real-world company, but each team uses the data to accomplish a different objective. For instance, for a recent project, students considered a company that sells mostly through retail distributors but planned to increase the importance of its online sales portal; the company’s leaders wanted to know what kind of website information and promotions would be more effective. Nine teams of students looked at different types of customers, from millennials to women.

As students progress through their courses, they learn information related directly to the project, says Cherie Speier-Pero. “When we covered marketing in our introductory class, we spent extra time talking about web analytics. In the statistics class, the professor incorporated homework assignments based on the company’s data, and students learned increasingly more sophisticated statistics techniques. We orchestrate the course assignments in the first semester to support the learning and activities happening in the project. Students learn quickly what their strengths are, where their gaps are, and how sophisticated their skills will need to be if they’re going to pursue data analytics as a career.”

Among the tools students learn to use at MSU is IBM’s SPSS modeler, a predictive analytics tool that “allows students to take existing data and make predictions about the future in a more robust and meaningful way,” says Speier-Pero. They also use the visualization tool Tableau to create visual displays from massive amounts of data.

“Databases and spreadsheets are still the tools of business, but we’re saying it’s not enough to show a pie chart or bar chart,” Speier-Pero says. “Visualization tools allow students and managers to drill down in the data right within the software. They allow users to explore data individually or in a group setting—say, if someone is making a presentation to the VP of marketing and a question comes up about sales. The tool allows users to explore data in a way that provides a more nuanced understanding and leads to better and faster decision making. I think visualization tools will soon be on every manager’s desktop.”

Speier-Pero notes that IBM and Tableau have made platforms available to students free of charge, provided training support, and attended academic conferences to help faculty understand how these tools can be incorporated into classes. Company reps come to class not only to provide technical training, she says, but to show students how to think about data using their specific tools.

In the fall, to draw students to the program before its January start, the university hosts an event called The Face of Analytics, in which about 50 companies come to campus to look for interns and full-time employees. Although about 35 students usually are enrolled in the program, about 300 show up for the event because they’re curious about job opportunities in the field. Says Speier-Pero, “Many get summer jobs where they learn what it’s like to be data scientists. Some go on to follow analytics careers, and some come back to join our program.”