Making artificial intelligence trustworthy and ethical

No one expects a keyboard, a hammer or a scalpel to have built-in ethical standards that guide its work. Only the users of such tools can chart those paths.

But artificial intelligence (AI) - the term given to the powerful computational tools that can capture enormous amounts of data, sort it out, learn from that data, apply it to problems, evaluate environments, make rapid decisions, create new imagery and written works, detect anomalies, solve problems and determine whether to accelerate or hit the brakes - is a whole 'nother tool.

It is, in fact, an amalgamation of data, tools, technologies, information and analytical processes, all connected in powerful and often undisclosed ways.

As its use expands and finds new applications around the world, the race to harness, steer and perhaps even master AI technology has been unfolding at breakneck speed. There is pervasive, consistent pressure to do more with it - as soon as possible.

At a recent symposium organized by the Data Science Institute at the University of Delaware, though, keynote speaker Katie Shilton made the case for a significant change in pace.

"We're in an era of 'hurry up and adopt,'" said Shilton, professor in the University of Maryland's College of Information and Institute for Advanced Computer Studies. "But we have to figure out how to support students and how this challenges fundamental science. We don't have to adopt fast."

Shilton's keynote was titled "No AI Without Us: Putting Co-Development at the Center of Ethical Data Science and AI." Co-development and participatory AI - the intentional inclusion of diverse voices and collaborators during the development of AI systems - were central concepts discussed. Shilton's Ethics and Values in Design Lab and the multi-university work of the Institute for Trustworthy AI in Law and Society (TRAILS), one of the National AI Institutes, are among the collaborative efforts working on these problems.