—James Truslow Adams, The Epic of American, 1931
Quoting Truslow in the introduction to his book, Reprogramming the American Dream: From Rural America to Silicon Valley—Making AI Serve Us All, Kevin Scott is unapologetically optimistic in his hope that perhaps more than any other technology that has come before, artificial intelligence has the potential to make that dream come true. Tech can serve humanity, he says, if we take the responsibility to make it do so. During a virtual CHM Live event on April 27, 2020, Scott, chief technology officer at Microsoft, discussed his ideas with long-time tech journalist John Markoff.
Scott grew up in rural Virginia, in a small, working-class town called Gladys. He spent time at the public library, where science fiction was a favorite, fascinating him with visions of future worlds that he wanted to be a part of. Like many, in order to fulfill that dream, he had to leave his hometown. He has a very personal understanding that broadband access to the internet is essential for rural (and urban) America to participate fully in a future where digital technology plays a critical role. COVID-19 is making it starkly clear that the digital divide is real: some have access to job opportunities and education online while others do not. Indeed, calls for Congress to fund broadband expansion are growing louder by the day.
A machine learning project he was involved in at Google in 2003 intrigued Scott with the possibilities of artificial intelligence. The field was finally beginning to move beyond what he felt were fairly dull rules and heuristics and into new realms where machines were being programmed to learn. Developments over the past couple of years take advantage of vast repositories of public data to feed into algorithms that can undertake a variety of tasks. These large-scale AI models, which used to be prohibitively expensive, says Scott, can now be available for people to solve new problems we can’t even imagine. But he has plenty of ideas about what those solutions might look like. Improving customer service and satisfaction is one. AI can be used effectively for rote tasks, reserving humans for those that require empathy.
There are plenty of applications for AI besides customer service. Scott is excited, for example, to see new advances at the intersection of biology and machine learning. He’s identified a recent pattern in scientific journals where researchers in the basic sciences are using AI in creative ways, a new calculus seemingly beyond traditional methods of modeling. Developments in AI for healthcare can create better outcomes for personal health and the healthcare system.
These dreams are within reach, Scott believes. But while progress is being made, it has plateaued in some areas and not advanced much in others, such as common sense reasoning or an artificial general intelligence. And how can AI have an emotional connection with humans? Or solve big problems like scarcity? As researchers work on these and other issues, Scott cautions certain principles must be adhered to in AI development, namely that it should be open, inclusive, human-centered, and beneficial. AI is not a magical entity and machine learning is not a mysterious process. They are tools we must choose to use wisely, accepting responsibility for their outcome and impact. We can all understand it and should not cede decisions about AI to “the experts.”
In response to questions from Markoff and the audience, Scott addressed potential promises and pitfalls of an AI future. Machine learning uses lots of data that comes from all of us as we navigate, click, and engage with technology, and AI can help businesses monetize that data. Scott explores the issue by channeling his 12-year-old self.
Bringing the conversation full-circle, Scott closed with a reminder that dedicated entrepreneurs working all over the country need help to turn their dreams into reality, not just access to broadband, which is fundamental, but also access to capital. In the past, help in the form of funding for research and development often came from significant investments by the federal government: the Cold War Apollo program seeded the entire aerospace industry; more recently, the DARPA Grand Challenge created an ecosystem for the development of self-driving car technologies. Scott suggests we should pick an ambitious problem to invest some percentage of GDP in solving, such as “ubiquitous high-quality healthcare for virtually zero cost to every person in the world.” This, he believes, would have huge implications not just for cost savings, but more importantly for improving quality of life and the flourishing of human potential. That would be an American Dream worth pursuing.