The smartest business leaders know that developing a strong artificial intelligence governance framework is not only the right thing to do, but it will also have a positive impact on their bottom line, especially as investors become more savvy and regulatory pressure increases.
Thoughtful leadership at the C-suite level is essential to deploying AI systems responsibly, according to Kans Kanka, director of responsible AI practices at the Institute for Experiential AI at Northeastern University and an associate research professor of philosophy.
“Leadership has to take responsibility and really drive this,” Kanka said, “because without them, this effort won't work.”
Kanka said this top-down approach can work across industries, from financial services companies looking to reduce the time it takes to analyze thousands of data sets to large retailers looking to automate customer service operations.
“It's not just ethicists and developers who are worried about what goes into AI systems,” she says.
Kanka was the instructor for the Responsible AI Executive Education course, held this week at Northeastern University's Boston campus. The two-day training was designed for current and future C-Suite executives who want to pioneer responsible AI practices within their organizations.
The education team included Ricardo Baeza Yates, research director at Northeastern University's Experiential AI Institute, AI ethicists Laura Haber Ehle and Matthew Sample, research scientist Annika Marie Schöne, postdoctoral researcher Muhammad Ali, and senior fellow Steve Johnson.
Johnson, CEO of Notable Systems Inc., which uses AI to help healthcare workers enter data, has developed his own AI framework in the form of five rules that he frequently shares with executives.
Rule 1: Don't believe in magic.
“If a vendor says, 'Just plug it in and it works; you go do something else,' that's probably not true,” Johnson says.
Companies may claim their systems are accurate 99% of the time, but what users should be concerned about is the 1% of the time the system breaks down, because that's when they might run into problems, he says.
Rule 2: Don’t be put off by the term “artificial intelligence.”
“I like to translate this to the word 'software,'” he says. “Proceed as you have with any software system in your career: Examine its complexities, its idiosyncrasies, its strengths and weaknesses. You always do that with any software system.”
Rules 3 and 4 go hand in hand: first, trust the system but make sure it continues to work correctly, and second, check with the vendor and make sure they have a system in place that allows users to verify the machine's capability to perform a given task.
Johnson gave the example of his own company, where he works in data entry, and the AI system he uses lets users know when it doesn't recognize certain information.
Rule number five is to ask the vendor how their system helps you close the loop.
“That means how the system gleans and learns from outcomes and from human intervention that occurs,” Johnson says. “The human element — your judgment and guidance — is a valuable source of truth. It's what keeps the system safe and healthy and improving.”
Andrew Glover, chief risk officer at Bangor Savings Bank, thought the meeting was a useful analysis of the issue.
He found Johnson's five rules an easy guide to follow, and Bangor Savings Bank wants to use AI to improve efficiency, he said.
“I always quote Jurassic Park,” he says. “I think Jeff Goldblum's character says, 'Just because you can, doesn't mean you should?' And I keep saying the same thing about AI: There's a lot we can do, but we always have to try. [ourselves] And do the right thing.”
