What policymakers should know about AI

Biostatistician Brian Caffo explains why Johns Hopkins University is hosting a briefing to help congressional staff learn more about artificial intelligence and its many applications

Few issues have caught the attention of policymakers like the future of artificial intelligence. With the potential to transform nearly every sector, lawmakers are now considering how—or if—to regulate AI.

Making those decisions relies on a deep understanding of the underlying technology and its applications, so Johns Hopkins University faculty members are hosting an information session on AI for congressional staffers. The goal of the session is simple: to connect policymakers with data and resources, as well as experts, that can help guide their decisions going forward.

Brian Caffo, a professor of biostatistics at the Bloomberg School of Public Health and the director of academic programs at the Johns Hopkins Data Science and AI Institute, is helping to organize the information session.

Here, Caffo talks about key policy issues surrounding AI, the upcoming session, and more.

How do you use AI tools in your research?

I work in biostatistics as it applies to neuroscience and neuroimaging work, where statistics, computing, math, neuroscience, and high dimensional measurement of the brain coalesce. In the process of doing that, I use a lot of tools, such as AI and machine learning, and because of that I’ve gotten involved in the new Data Science and AI Institute.

I’ve always been more of a machine-learning researcher, but in my neuroscience work, we spend a lot of time studying things like intelligence and learning. So artificial models of intelligence come into play there, though in a different way.

What can you share about the upcoming information session?

As part of our academic programs and education effort, we want to perform what I would describe as a public service—education. The idea here is to match congressional staffers with top researchers in the field of artificial intelligence that come at it from several different angles.

We have a lineup of people that includes some people who use artificial intelligence more like I do in the pursuit of biological, medical, and public health knowledge. We have some people who focus on language research and models, and others who study image and pattern recognition. The event also includes policy and ethics experts.

I’m especially interested in a colleague of mine from Biostatistics, Liz Chin, who is going to talk about AI and ethics. She’s worked on things related to AI in the carceral system. I’m more of a technical person; I tend to focus on the way I use AI for diagnostic images. So I’m excited because it’s a little different than what I work on, but also super relevant and very important.

Mark Dredze will be talking about AI for speech and language, things like ChatGPT. He’s long been an expert in AI for language and is an incredibly fun and entertaining presenter.

I also think it’s always fun to hear my colleague Haris Sair talk about radiology. It’s certainly the medical field where AI is having the biggest impact. It’s interesting to see how somebody at the forefront of this talks about it.

What do you think are the top AI policy issues?

By comparing the U.S. to different countries, you can see different policy agendas are starting to materialize. There’s a wide range of issues.

A classic example is the ethical use of AI in sentencing, determining bail and things like that. The idea of how one fairly utilizes AI, so that it doesn’t reflect the biases that exist in the data itself, is a challenging problem. Certainly, AI and intellectual property is a core component to be discussed. 

AI policy is a high-dimensional problem that requires many different angles and opinions. That’s why organizing a session like this, which includes researchers using AI in many different fields, is quite useful.