The University of Maryland, Morgan State University and George Washington University are involved in a $140 million research and development project involving artificial intelligence systems.
The idea behind the National Science Foundation’s Institute for Trustworthy AI in Law and Society, or TRAILS, is to broaden who develops AI, study how people use technology in their everyday lives, and increase public trust in those systems.
“AI is a big part of our everyday lives,” said Virginia Byrne, assistant professor in the Higher Education Student Affairs Program at Morgan State University. “AI backs all of those great social media apps we love to scroll through.”
Byrne said that, as the tech becomes increasingly sophisticated and employed in programs people use every day, broadening who develops it and increasing the understanding of how it can be used is more important than ever.
She’ll be working with students, teachers and teachers who are in “pre-service,” those heading into the field of education, on the TRAILS project.
“We’re going to work with these three groups to explore how they feel about AI, how they use AI in their everyday lives and the sticky issues around trust with AI,” Byrne said.
Hal Daume, computer science professor at the University of Maryland, which will colead TRAILS, said the issue of trust cuts both ways.
There are times when people over-rely on AI, according to Daume.
“I’ve seen people use chatbots to get medical advice. That is probably not a good idea,” he said. “You probably shouldn’t trust a random person on the web to give you medical advice, you probably shouldn’t trust these models to give you medical advice.”
At the same time, Daume said mistrust of AI can keep people from using it in ways that can help them tackle mundane tasks.
Part of the work of the TRAILS institute will also focus on how people — including students from K-12 and beyond — actually use AI in the real world.
Daume, who’s teaching an undergrad AI course this semester, said he had a speaker come to his class, and during her presentation, Daume sat among his students.
“I saw students on their laptops who were pulling up ChatGPT to work on homework for other people’s classes, which I thought was kind of a bold move knowing that your professor is sitting behind you.”
But he said the experience made him realize that a lot of faculty — including himself — don’t fully understand how students are using technology today, something the TRAILS research is intended to address.
While discussions of AI can prompt concerns that robots are ever closer to taking our jobs, Daume said having more types of people involved in the development of AI — another objective of TRAILS — can alleviate those concerns. There is no doubt that technology has changed the labor landscape, but Daume said the focus should not be on building systems that replace human labor, but on programs that can remove the tedium from some jobs.
When asked about how AI could help people do their jobs, Daume said “The answer is, in my experience, never ‘Do X for me’, the answer is always ‘Help me with these tedious parts of doing X so that I can spend my time doing the interesting parts of doing X.”
The latest $140 million investment involved a total of seven new AI institutes.
