Juan M. Lavista Ferres got his start with AI for good in an unlikely way: He really didn’t want to go hiking.

While working at Microsoft running randomized control experiments, Ferres had a colleague who was raising money for research on sudden infant death syndrome, or SIDS, to which he had lost a child. The campaign involved climbing the Kilimanjaro mountain — decidedly a no-go for Ferres.

Ferres didn’t want to climb a mountain, but he did want to contribute to his friend’s cause. So rather than lacing up his hiking boots, he got together with a group of colleagues to analyze the U.S. Centers for Disease Control and Prevention’s cohort-linked birth/infant death dataset, an open-source dataset that’s logged tens of thousands of SIDS cases in the United States.

Using machine learning models, Ferres and his colleagues were able to manipulate the vast amount of data in ways that had never been done before, revealing new links between risk factors in infants and SIDS. It was through this effort that Ferres realized how powerful artificial intelligence could be for solving some of the world’s most intractable problems.

For many of us, AI means asking ChatGPT for recipe suggestions or, more darkly, ruminating on the possibility that the machines will soon eclipse human intelligence. But “AI for Good,” both the name of Ferres’s book and the lab he directs at Microsoft, demonstrate the enormous power of AI to improve the world as we know it.

In this episode of the Devex Book Club podcast, President and Editor-in-Chief Raj Kumar sits down with Ferres to talk about how he got his start with computer science, what AI can do (and what it can’t), and how beluga whales can help detect war crimes in Syria.