Amidst the benefits that algorithmic decision-making and artificial intelligence bring – including revolutionising speed, efficiency and predictability in many sectors – Manish Raghavan is exploring opportunities to apply these technologies to address existing societal concerns while also mitigating the associated risks.
“Ultimately, I hope my research will lead to better solutions to society’s long-standing problems,” says Raghavan, the Drew Houston Career Development Professor and joint faculty member in the Department of Electrical Engineering and Computer Science at the MIT Sloan School of Management and the MIT Schwarzman College of Computing, and principal investigator in the Laboratory for Information and Decision Systems (LIDS).
A prime example of Raghavan’s intent can be seen in his research on the use of AI in recruiting.
“It’s hard to argue that past hiring practices are particularly good or worth maintaining. Tools that learn from historical data will inherit all the biases and mistakes humans have made in the past,” Raghavan says.
But here Raghavan raised a potential opportunity.
“Discrimination is always difficult to measure,” he said, adding, “AI-driven systems are sometimes easier to observe and measure than humans, and one of the goals of my research is to understand how we can leverage this increased visibility to find new ways to detect system malfunctions.”
Raghavan, who grew up in the San Francisco Bay Area with parents who both have degrees in computer science, says he initially wanted to be a doctor. But even before he enrolled in college, he discovered he loved math and computers, and wanted to follow in his family’s footsteps and pursue a career in computer science. After spending a summer as an undergraduate working with John Kleinberg, a professor of computer and information science at Cornell University, he decided he wanted to pursue a PhD there and wrote his thesis, “Social Implications of Algorithmic Decision-Making.”
Raghavan has received numerous awards for his research, including a National Science Foundation Graduate Research Fellowship, a Microsoft Research Doctoral Fellowship, and a Cornell University Computer Science Department Doctoral Dissertation Award.
He joined the MIT faculty in 2022.
Perhaps inspired by his early interest in medicine, Raghavan investigated whether the discrimination outcomes of a highly accurate algorithmic screening tool used to triage patients with gastrointestinal bleeding, known as the Glasgow Blatchford Score (GBS), could be improved by adding expert advice.
“GBS is, on average, as good as humans, but that doesn’t mean there aren’t individual patients or small groups of patients where GBS might be wrong and the doctor’s diagnosis might be correct,” he said. “If we can identify these patients in advance, our hope is that doctor feedback will be especially valuable.”
Raghavan also studies how online platforms influence users, noting how social media algorithms observe the content users select and show them more of the same. The challenge, Raghavan says, is that users can choose what to watch in the same way they choose a bag of potato chips. Potato chips are tasty, of course, but they’re not all that nutritious. The experience may be momentarily satisfying, but it can leave users feeling a bit queasy.
Raghavan and his colleagues developed a model that shows how users with competing desires for immediate and long-term gratification interact with platforms. The model shows how platforms can be redesigned to promote healthier experiences. The model won the Exempary Application Model Track Paper Award at the 2022 Association for Computing Machinery Conference on Economics and Computing.
“Even if you only care about the company’s profits, long-term satisfaction is what matters most,” Raghavan says. “If we can build up evidence that the interests of users and companies are more aligned, then hopefully we can promote healthier platforms without having to deal with conflicts of interest between users and platforms.” Of course, this would be ideal. But I feel like there are enough people in these companies who believe there is room to make people happy, but they just lack the conceptual and technological tools to make it happen.”
When it comes to the process of coming up with ideas for these tools and concepts that most effectively apply computational techniques, Raghavan says his best ideas come when he thinks about a problem for a while. He said he advises his students to follow his own example of taking a really difficult problem, putting it aside for a day, and then coming back to it.
“The next day things usually get better,” he said.
When he’s not solving problems or leading, Raghavan is off the soccer field coaching the Harvard University Men’s Soccer Club, a job he cherishes.
“Knowing I’ll be out in the field until the evening keeps me from procrastinating and gives me something to look forward to at the end of the day,” he said. “I try to build things into my schedule that I consider to be just as important as the work itself, to help me put challenges and failures into context.”
When thinking about how computational technology can contribute to our world, Raghavan says what excites him most about his field is the idea that AI will open up new understanding of “humans and human societies.”
“Hopefully we can use this to understand ourselves better,” he said.
