Artificial intelligence (AI) systems are continuing their rapid ascent into almost every aspect of our lives. How—and how well—these systems interact with humans offers a trove of thought-provoking questions. Jordan Boyd-Graber, a professor of computer science at the University of Maryland Institute for Advanced Computer Studies and the Institute for Trustworthy AI in Law & Society (TRAILS), is studying the current challenges involving AI-infused systems, and how they may soon significantly enhance humanity, not replace it.
Boyd-Graber would not have recommended using chatbots for anything four years ago, but says that huge advancements have been made in the last few years. Despite those advancements, there are still big problems with these systems, like how confident they are about wrong answers.
“Not only do they not know when they’re wrong, they don’t know when they’re right. And usually if you ask them about their confidence, they’re very confident no matter what,” says Boyd-Graber, cautioning against the use of these systems in high-stakes applications.
Looking ahead, Boyd-Graber is excited about a world where “we optimize humans and computers working together, rather than trying to replace humans with AI.”
—Produced by UMIACS communications group