Relax, Northeastern’s first AI chief is a human being

Javed Aslam, a computer science professor with more than 20 years of experience at Northeastern, was named the inaugural chief of artificial intelligence.

Portrait of Javed Aslam on an orange background.
Photo by Matthew Modoono/Northeastern University

Javed Aslam, a computer science professor with more than 20 years of experience at Northeastern, was recently named the university’s inaugural chief of artificial intelligence.   

David Madigan, the provost and senior vice president of academic affairs, made the announcement recently. 

“With extensive AI experience in both academia and industry, Jay is well-positioned to excel in this new role. As a professor in the Khoury College of Computer Sciences, his research has focused on AI and its applications, including machine learning, information retrieval, natural language processing, and healthcare,” Madign said in the announcement.  

Aslam also serves as the chief data science officer at CodaMetrix, an AI-healthcare company based in Boston that uses AI tools to help clients automate some of their medical coding processes. 

Aslam has been working at Northeastern since 2003 and he has seen it evolve over the past 21 years. Northeastern Global News caught up with Aslam to ask him about his new role and his time at the university. His comments have been edited for clarity and brevity. 

Provost Madigan said in your new role you’re going “to promote AI innovation, literacy, and adoption across research, educational programs, and institutional operations.” In your own words, how would you describe the job? 

Innovation. I think innovation is all about how we can move forward and advance the field. We already have a strong set of researchers in core AI methodologies, AI policy, responsible AI and other areas. One of the things we want to do is build on those initiatives to lead in those areas through targeted hiring and providing the resources that are necessary for those researchers to thrive. 

For example, one of the major areas for AI these days is large language models that underpin things like ChatGPT. But one of the issues in doing core research in that area is that it requires significant computing power, and a very specific kind of compute that requires GPUs (graphics processing units). One of our initiatives is to really outline a university-level strategy for the computing power that is necessary to perform this cutting-edge research. 

There’s also AI and research, specifically. We really want to leverage our strengths in interdisciplinary and use-inspired research. The interdisciplinary faculty who are working on core AI methodologies and AI policies and so on, are already doing the work, but there’s many other places where AI can actually supercharge the research of others. … I think part of it is both a data strategy and a model strategy. There’s a lot of data that’s out there that can help the researchers. How can we get our hands on that data and build AI-based models that are built on top of that data to help research?

I think the workers of the future are going to have to be literate in AI, its capabilities, and the places where it’s not so capable. So how can we create a core AI course that could promote AI literacy for students across the entire university? How could we create AI minors?  How can we help educate our faculty and staff on the use of AI?… I think there is a lot we could do at the university-level to create and promote literacy within AI through courses, minors, certificates and so on.   

Why does Northeastern need a chief of AI? 

First, there are a lot of good efforts that are going on but many are happening in isolation across the university. … One of the reasons to have this role is to help unify and strengthen these efforts so that they are more likely to succeed and have impact. There is a person who can help manage all these efforts. 

There’s my role, but we also have an AI Council at the university drawn from leaders across the institution. We’re going to have work streams led by university experts that correspond to all the efforts that I described earlier. You need someone who can manage these efforts with a unified strategy for how we can push forward everywhere. 

Can you share how your experience as an AI researcher and professor has prepared you for this role? 

In some sense, I’ve run the gamut all the way from the theoretical end of machine learning through to its practical applications. I think that’s incredibly useful in terms of having a broad view. I’ve been involved in machine learning for decades, and I’ve lived through its boom- and-bust cycles. There were times when people thought AI and machine learning were going to take over the world and then it didn’t fulfill its promise and it crashed. 

I think we are certainly at a rise at the moment with the advent of large language models like ChatGPT. People think it’s going to do amazing things, but I think it’s really important to not only understand the hype, but to really deeply understand what these things are capable of, what they are not capable of, and why. If you know that you can really make use of these things profitably. 

How will your industry experience, primarily serving as the chief data officer at CodaMetrix, help guide you? 

It really gets back to one of our focuses at the university — use-inspired research. In some sense, that has been my role at this company. We have a particular use case in health care that has inspired us to build these AI technologies. 

I think working in industry where you are confronted with real-world problems and real-world data has been incredibly enlightening to me as a researcher and a professor. For example, if you were teaching an academic course in machine learning, you would often work with standard benchmark data sets that tend to be reasonably noise-free. But in the real world, data is oftentimes really messy.  It’s an eye-opening thing to experience and confront. 

If you want to make use of your AI, you really have to have a good data strategy, a good data-cleaning strategy, a way of handling missing values and features, noise, and the like. This is what it means to apply AI in the real world. That should inform us on the academic side.