In his last role at Microsoft, John Kahan headed up what would become Microsoft’s AI for Good efforts. To this day, John continues to use data and AI to help solve the world’s problems whether it’s via his work with Aaron Matthew SIDS Research Foundation at Seattle Children’s hospital or advising organizations that include U.S. Venture, MidOcean Partners, Stagwell, Novartis Foundation and the Institute of Health Metrics and Evaluation. In this episode of Beyond the Blue Badge, host Rich Kaplan talks with John about advances in data science, AI for Good and the evolution of artificial intelligence.

The start of AI for Good: John worked with Microsoft President Brad Smith and the senior leadership team to launch AI for Good, an effort to use data and AI to help solve world problems. “It really goes to the heart of Satya (Nadella)’s strategy to empower and enable everyone on the planet, not just those that have money, but those that require the use of technology to be able to help solve some of these challenges.”

Starting with SIDS: John’s fourth child, Aaron Matthew, died of Sudden Infant Death Syndrome (SIDS), the leading cause of death for infants one month to one year old. He and his team of data scientists with the support of Microsoft went to work. “We had launched an initiative with Seattle Children’s to help stop SIDS,” he said. “And our data scientists and our high-performance computing with Azure, they all leaned in and an incredible miracle happened. Today, we understand 26% of all SIDS deaths worldwide…we now understand why my son died, as an example.”

Expanding the mission: Brad, Satya and other Microsoft leaders encouraged John to put the same techniques to work on other major problems including cultural heritage, environmental sustainability, needs of children and humanitarian action. “I got an opportunity over about a three-year period to be able to use my talent and skills along with others across Microsoft and our technology to help lean in and solve some of these challenges,” John said. “It was an incredible mission and really a lot of fun.”

Dearth of data scientists: At the time that AI for Good started, they discovered that more than 50% of data scientists were working in high tech and fintech. Fewer than 5% were working with nonprofits or in health care. The epidemiologists at Seattle Children’s had a mountain of data, but not the ability to look for patterns, causation and correlation. Microsoft’s data scientists and technology could help solve that problem. “When you realize that we have basically starved the researchers in areas that really change lives, you have to give back,” John said. “So, you marry the skills we have — high-performance computing and the data that the U.S. government had there all of the time in plain sight — and you start to realize you can connect these dots all together.”

AI for Health: Right before the COVID pandemic occurred, Microsoft launched AI for Health. During that time, more than 1000 Microsoft employees across 100 different countries contributed data and insights to the data science teams to help understand the spread of the disease and what was occurring and where it was occurring to help protect people not just at Microsoft, but their families, friends and the communities that are served around the world. “We never could have gotten the progress we had made during Covid if we didn't have Microsoft employees volunteering from across the world to be able to feed and fuel the efforts that occur,” John said.

AI is old-school: John points out that the idea of artificial intelligence has been around since the 1940s. But generative AI has “woken up the world” to what data scientists have been doing for a long period of time. “It’s moving at a pace and a speed that we have never seen in the years that I’ve been working in this space,” John said. “What makes it incredibly exciting is that we’re opening up the tools and the capabilities to the non-data scientists, to anyone that has a browser or an application or a computer for the first time.”

Modeling ethics and values: With adoption of generative AI happening exponentially, John points out real concerns of making sure that the technology is built responsibly. We need to build diverse models with diverse teams and capabilities. Companies need to build responsible AI directly into the products and services, and they need guardrails. “Certainly, with something this powerful, we will make mistakes,” John said, but noting that we need to learn from those mistakes. “We will hopefully fail fast, correct and adapt and move forward.”

Protecting privacy: John walks us through the way generative AI models learn and how companies are protecting personal data.

Bridging a gap: John’s tech career spanned 40 years at IBM and Microsoft, but he took the opportunity to work with other industries via his board service. He runs the heard I run a foundation called the Aaron Matthew Research Foundation at Seattle Children’s, which uses AI and data to help solve sudden infant death syndrome. He works with the Institute of Health Metrics and Evaluation out of the University of Washington, a Bill Gates-funded organization that identifies health measures that are used by the World Health Organization and others. He’s also on the Novartis Foundation Board, using AI to help solve cardiovascular disease in low-income areas around the world.

Re-learning to be a human: We’ll all need to re-learn how to ask computers for information. Large language models will let people use normal conversation to procure information.

Photo courtesy of John Kahan

John Kahan spent more than 40 years at IBM and Microsoft and helped launch Microsoft’s AI for Good program. Now, he’s putting his data science background to work advising a number of nonprofit and for-profit organizations.