Throughout June 2023 we are showcasing interdisciplinary artificial intelligence (AI) research at the U of A that demonstrates how the university is leading with purpose to make AI safer, reliable and more just.
Eleni Stroulia, professor in the Department of Computing Science and Vice Dean of the Faculty of Science, works on ways of using AI to inform decision making in many fields.
In this week’s spotlight, Eleni speaks to the importance of combining reinforcement learning, and more generally data-driven machine learning, with expert knowledge and the necessity of understanding data and computational thinking.
What is AI?
There are many definitions of AI and, especially today, we frequently see a definition based on data driven AI algorithms, but my favorite definition comes from 1955. John McCarthy, who was a professor at Stanford and is often credited with being the father of AI, defined AI as the science and engineering of building intelligent machines. That definition covers both aspects of science and algorithm development, and also the necessary implication that we have to build these machines and make them work in the real world.
Briefly explain your field of research and how it involves AI.
My PhD was in artificial intelligence and my thesis was on how we can learn from negative examples. Intuitively, when we fail as humans, there's something to learn in order not to fail again in the same way in the future. Learning from negative examples enables us to compare what we know should have happened and what we know went wrong, and figure out how these differences can guide decision making ahead of time. Building on that, in my research I have been working on ways of using AI to inform decision making in many fields. I am working with many researchers across campus, conducting interdisciplinary work on how these kinds of decision making algorithms based on heuristics, search, and data-driven analysis can help with health, engineering, construction and education. It has been a pleasure to see all this work have an impact, smaller and large, on real world activities. I have also been working with industry using AI to develop better software development processes, so that the software that is deployed is less buggy and more efficient.
How is AI affecting our lives and what is a common misconception people have?
It's difficult to talk about AI today and not think about ChatGPT. ChatGPT is a very important advancement demonstrating the power of data and is truly impacting our lives, especially as instructors. We have to figure out how we should change our pedagogy to make sure that our students, who are likely to use ChatGPT, take advantage of this technology and they're not simply using it to not do the work we assign to them. But ChatGPT is a very small slice of what AI work is about. AI covers the areas of robotics, industrial automation and controlling cyber physical systems, anywhere from buildings to cars to cities. And it's very important for people to not lose sight of all these opportunities and all this potential just because of this important advancement that is grabbing our attention today.
What is the long term future of AI? And how is the U of A leading in this space?
There are two broad concepts of artificial intelligence and machine learning today. One is this notion that data can be analyzed and generate evidence that can guide decision making, which is a very popular area of activity today. The other, which is not as popular today but is where expert systems started, is this notion and belief that the expert knowledge and the skills of enlightened humans should be embedded in algorithms.
I think that our challenge today, and the work that we should be doing for the long term, is to merge these two concepts and these two types of work. Learning from data is absolutely essential, especially since software and hardware today can give us a wealth of data from which to learn. But it's also important to start from a level of knowledge that takes advantage of people's expertise, not always start from first principles.
The U of A is the stronghold of reinforcement learning. This is our particular area of expertise and it is important to understand that reinforcement learning merges data and expertise. Agents trained with reinforcement learning learn from their experiences as they're given feedback that can incorporate expert knowledge. This is one of the areas that holds much promise for the future, and there is much more work to be done. I believe the U of A has a huge role to play in this.
What do people entering the workforce need to know about AI?
I think everybody today has to be informed about AI because there is so much noise. We are listening to stories about how AI is taking away people's jobs or preventing people from being hired into the jobs they want, and that is, to some extent, true. There are two things we need to understand as we're making decisions about careers and about our own work in the field.
First, everybody has to be informed about data and the opportunities around data. I believe that even our high school education should take into account basic data analysis methods with statistics and relevant computational tools.
The second aspect of our education should involve an understanding of computational thinking. Essentially, we all need to be trained in decomposing problems, abstracting them to problems that we have already solved, being able to recognize patterns and being able to apply these patterns in our decision making. These are fundamental skills that everyone should be trained in. And we should all be questioning the information that we're getting, trying to find our sources and make sure we understand what is true, what is applicable and what is not.
As citizens, we should be asking for legal frameworks to be placed around AI systems and more generally, software systems that are embedded in government decision making. Commercial software takes our data and uses it for multiple purposes. We should all be making sure that these activities are both allowed and amplified, but also contained into proper laws that take care of people.
This conversation has been edited for brevity and clarity.
About Eleni
Dr. Eleni Stroulia is a Professor in the Department of Computing Science at the University of Alberta. Her research focuses on addressing industry-driven problems and interdisciplinary challenges, using AI and machine-learning methods. From 2011-2016, she held the NSERC/AITF Industrial Research Chair on Service Systems Management with IBM, and she has played leadership roles in multiple national and international projects. She has supervised more than 60 graduate students and PDFs, who have gone forward to stellar academic and industrial careers. Her excellence in mentorship has been recognized with a McCalla professorship (2018) and a Killam Award for Excellence in Mentoring (2019). From 2020 to 2023, she led the University of Alberta's AI4Society Signature Area, and has served as the Vice Dean of the Faculty of Science since 2021.
Innovator Spotlight is a series that introduces you to a faculty or staff member whose discoveries, knowledge and ideas are driving innovation.
Do you know someone at the U of A who is transforming ideas into remarkable realities? Maybe it’s you! We are interested in hearing from people who are helping shape the future, improving quality of life, driving economic growth and diversification and serving the public. We feature people working across all disciplines, whether they are accelerating solutions in energy, shaping the evolving landscape of artificial intelligence or forging new paths in health and Indigenous leadership.
Get in touch at blog@ualberta.ca.