A national leader in artificial intelligence returns from DC on a mission to elevate UT as a leader in the field.
Lynne Parker recently returned to the University of Tennessee, Knoxville, after completing a four-year post as deputy US chief technology officer and founding director of the National Artificial Intelligence Initiative Office within the White House Office of Science and Technology Policy.
Parker now serves as associate vice chancellor and director of the new AI Tennessee Initiative at UT. Drawing on her experience at the White House—where she oversaw development and implementation of the national artificial intelligence strategy—and nearly 30 years of robotics research as a professor at UT and researcher at Oak Ridge National Laboratory, Parker will lead the university’s strategic vision for multidisciplinary AI education and research.
The AI Tennessee Initiative at UT is designed to increase the university’s funded research, expand the number of students developing interdisciplinary skills and competencies related to AI, and position UT and the state of Tennessee as national and global leaders in the data-intensive knowledge economy.
We sat down with Parker to learn more about AI and what she hopes to bring to UT through this new initiative.
What is meant by artificial intelligence?
Generally speaking, we think about artificial intelligence as software or other types of technologies that perform behaviors very similar to what we would expect humans to be able to do, like perception, cognition, and action.
Artificial intelligence shows up in many applications of everyday use. This includes natural language processing. For instance, when you call to get help for some sort of service and you have a chatbot that answers the phone and you talk with that system, that is an AI system. Or you may go to the doctor for a radiology scan and the doctor may use an AI tool to help define suspicious areas that are in that scan to help better diagnose you.
What do you see as new frontiers for AI research?
Current AI systems are based on large-scale data, but humans don’t require large-scale data. I like to give the example of a child looking through a picture book, and the child sees a picture of a fish. After just a couple of looks at that book, the child and their parents can go to an aquarium and see a real fish. The child will just say fish, because they’ve learned from a few examples and a picture book that this is a fish.
AI systems don’t have the ability today to learn from just a few examples. So this area of data-efficient learning is an area in which I think there is tremendous potential for increased understanding, increased knowledge, and increased advances in AI systems.
A second area is in commonsense reasoning. People are very good at this. For instance, you would not put a stack of heavy books on a child’s play table, because you know a play table cannot hold a stack of heavy books. Commonsense reasoning is very difficult to mimic and is not easy to program into an AI system. There are many researchers who are studying this. It’s not an easy challenge but one I believe will have significant advances going forward.
There are definitely many types of advances yet to come for AI systems. In five to 10 years, I believe that AI systems will begin to require much less data. One of the benefits is that it will be easier to develop applications for certain types of systems because we won’t require as much data that can be hard to develop. Not having this data also means we won’t necessarily have the privacy or security concerns that we have with much of the data today.
Can you tell us more about the AI Initiative at UT?
Nearly every country has an AI strategy. So it has, in these last five years, really become a key area that people know is going to impact our lives and in all ways and all sectors.
Here at UT, I am looking at the research and the education initiatives in particular, and ways we can leverage the unique strengths of Tennessee and the University of Tennessee to engage in all the efforts happening now across the nation. I’m a big believer in how we can ensure that all parts of the country can participate in this great growth in AI.
We certainly want the University of Tennessee to be a leader, but we want our nation to be a leader as well. We can do that by showing an example of how it can be done, building up the faculty expertise, building up the infrastructure we need to be successful, building up the curricula across education programs, building the partnerships across the state that can make us successful, and demonstrating a model that can be used to help the nation as a whole build up that innovation ecosystem, especially across the interior of the country.
How can UT students get involved?
Students at all levels will benefit from this and will be able to participate in this initiative. AI is being used in every domain. It’s not just a research tool; it is a tool to solve problems. We want all our undergraduates, as they go out to whatever their next steps are—be it industry or the military or graduate school—to have an understanding of AI.
It’s not just about computer science. For example, if you’re an English major, you might understand how AI systems are now being used to come up with first drafts of documents and how that should be used, what that means for the creative process, and how it can be a tool to maybe help those with writer’s block get started writing about something.
We have a shortage right now across the country of talented people who can work in this space with multidisciplinary perspectives. There are certain leaders on campus in different disciplines that are already using AI and are known for their leadership in this field.
When people became aware of this initiative, I had an outpouring of emails reaching out to me saying “This is fantastic, how can I get involved?” I have some ideas, and I’m really enthusiastic about talking with people across campus and finding out what their ideas are, and how we can leverage those unique opportunities and partnerships.
Can you address the need for diversity and inclusion in AI?
AI is a multidisciplinary field that is being applied in nearly every sector of society. It automatically requires us to have a diverse and inclusive set of participants and stakeholders who are involved in the development, the design, and the use of these systems. By its nature, to be effective, it must be diverse and inclusive.
There have been a lot of concerns about AI systems that are inequitable, meaning the data being used in these systems can lead to certain types of bias or lead to propagation of stereotypes in the human world that we don’t want to mimic or continue in the AI system.
So we are beginning to learn more about how to overcome this type of bias and stereotype. Over time, we will continue to get better at that. In addition, there are many types of situations in which AI systems can detect bias in human systems and alert people to that so that the overall system can improve and that humans can be aware of the types of biases that may be at play and to overcome those.
What is it like for you to come home to UT?
What better place to come home to than a wonderful university that is ready and primed? There’s a lot of enthusiasm, there’s a lot of excitement about building up these capabilities—not only on our campus here in Knoxville but across the system and across the state—with partnerships that we can form with industry, with other universities, with civil society. All of this is designed to help humans make life easier, which goes in line with the Volunteer spirit.
It’s good to have had the opportunity to work in the policy world, to see the overarching things that need to be done to make sure the nation has the right priorities as well as the right emphasis and strategy. But at the end of the day, we have to actually implement these strategies. That’s what I’m excited about doing here at the University of Tennessee.
We have amazing things happening here. We have great leadership and vision. We have amazing faculty and staff. We have fantastic students. And we have top-notch programs in so many different fields of study.
AI systems are, in my mind, always designed as a tool to work with people. When you think about a tool that helps people do their jobs better, there is an interesting correlation to the Volunteer Creed and our Volunteer spirit of trying to be helpful.
AI systems are being designed with this very idea in mind. They’re not taking over, but helping us to do our jobs better, making sure we’re overcoming tough challenges, and helping all sectors of society do good.
Christie Kennedy (865-974-8674, firstname.lastname@example.org)