When a classic action movie appears on your Netflix home screen or a new mystery series rises to the top of your Amazon account as something you might like, you know it's an algorithm at work. But those are just the transparent ones.
Increasingly, algorithms are everywhere, for they are the digital DNA of modern life. They aren't just making media suggestions. They're also enabling every online search, screening job candidates, powering customer service chatbots, and deciding who qualifies for a bank loan.
“Algorithms and data-driven decision-making are being used more frequently to power all kinds of decisions within organizations,” says Omar Besbes, the Vikram S. Pandit Professor of Business. The study of data-driven algorithmic decision-making is at the heart of a new CBS think tank helmed by Besbes, the Algorithmic Economy Lab, which is part of the School's Digital Future Initiative.
Given the widespread adoption of algorithms across all types of industries, it's imperative that we research and refine this still-evolving field, Besbes says.
“There are a variety of questions that arise about the technology that supports real-time decision-making with data-driven algorithms,” he adds. “But there are also questions about the value that can be captured through more sophisticated algorithms—and the sometimes unintended consequences that can emerge as a result.”
The new lab will delve into these and other unanswered questions by focusing on research, teaching, and engagement with the business community. One key area of investigation will be the issues that may arise from the proliferation of algorithms, and why it's important for us to understand how they interact.
For instance, if two competing firms use similar but different algorithms, what are the implications? Will the outcomes be more accurate? Or might the results influence each other and lead to less accurate outcomes? If one firm's algorithmic trading orders are triggered by a specific set of circumstances, such as having a preset buy order for a target price-earnings ratio, how might the market be influenced by another firm's preset order? With the potential for complementary or competing algorithms triggering trades, what are the unintended consequences and potential risk exposure for the firms? Promoting a mindset of algorithms as a system helps unearth these essential questions, says Besbes.
Another critical area to explore is trust in the way algorithms are used. Besbes notes that customers are more willing to use services where trust and transparency are central. The lab will therefore explore how to avoid biases, ensure privacy and transparency, and increase trust in an algorithmic economy.
Preparing students to successfully parse data— and manage teams and companies that rely heavily on algorithms—requires a parallel shift in the classroom. Even a decade ago, companies appreciated that data analytics was critical, but the focus was on hiring data scientists, Besbes says. Over time, there has been a growing understanding that data scientists may lack the institutional knowledge to ask the right questions, while those who ask the right questions may not always have the technical abilities to answer them.
The CBS curriculum teaches students to understand data science and have the business acumen to help a company solve the problem at hand, Besbes says.
“There's a sweet spot where individuals are digitally and technically literate but not necessarily specialists—and they can challenge data scientists,” Besbes says. “Columbia Business School's graduates are uniquely positioned to play that role because with their business acumen, they're able to ask the right questions but also have the sufficient technical and digital literacy to be able to challenge and guide the analysis.
