University of South Carolina researchers are working to shape how future engineers will work with AI by combining real-world engineering challenges with studies on ethical AI literacy.
The collaboration between the Molinaroli College of Engineering and Computing and the College of Education will develop and test new ways to help students decide when AI is useful, when it should be questioned and how its use can affect fairness, transparency, privacy and the broader society.
Mechanical Engineering Professor Yi Wang and College of Education Associate Professor Hengtao Tang have received National Science Foundation funding for the three-year project. Together, they will develop and test new ways to help students decide when AI is useful, when it should be questioned and how its use can affect fairness, transparency, privacy and the broader society.
“We knew we needed to come up with something to make sure AI literacy could be emphasized and understand how students make decisions when they start to use AI,” Wang says. “The appropriate use of AI and its value for research and education is most important.”
The collaboration places engineering and education research on equal footing. Wang will focus on creating realistic engineering scenarios in areas such as robotics, drones, computer vision and human-robot collaboration. Tang will lead efforts to define ethical AI literacy, establish practical principles and develop methods for assessing how well students understand and apply those principles.
For Tang, the challenge is turning broad discussions about responsible AI into something students can put into practice.
“The ethics of AI is about fairness, transparency, privacy and societal impact,” Tang says. “But how can we embed those principles into teaching and learning practices? That’s where there is a gap in translating theory and concepts into practice, and that’s something we want to address.”
Rather than simply giving students a list of rules for using AI, Wang and Tang want to place them in realistic situations where they must decide for themselves how — and whether — to use it. For example, a student working with a robot might have to decide whether to rely entirely on AI to provide a solution or use it as a guide while working through the problem independently.
The ethics of AI is about fairness, transparency, privacy and societal impact.
“That’s the moment that the student needs to make a decision: ‘Should I completely rely on AI to give me the solution for operating robotics? Or should I work on it and let AI guide it?’” Wang says. “This is how they can have an in-person experience with when and how to use AI, and it’s better than just providing rules and principles.”
Wang and Tang have already developed an open, intelligent robotics platform they hope to expand into course materials that can be shared with engineering students. Additional scenarios involving drones, computer vision and human-robot collaboration are planned. Working with those real-world applications is important because ethical questions may not become apparent until students are faced with making decisions.
“Only when you work on something real and practical can the problem be exposed,” Wang says.
The researchers also plan to develop short online or hybrid learning modules in which students evaluate AI or AI-enabled robots and recommend guidelines for their safe and responsible use. Researchers will be able to observe how students approach those situations, including when they rely on AI and when they choose to work through a problem themselves.
Tang's education research will provide the other half of the project.
One of his primary tasks is developing a framework that clearly defines ethical AI literacy and how it can be incorporated into engineering education. The researchers will work with experts in AI ethics, engineering education and assessment to identify a core set of concepts that students should understand. Tang will also lead development of a new assessment method to determine how students connect and apply those concepts.
Traditional approaches to assessing ethics or AI literacy often rely on surveys and interviews in which students describe what they know. But Wang and Tang plan to use student-generated causal maps, which are visual diagrams showing how students connect different ideas, factors and potential consequences. The maps could provide researchers with a window into students' thinking as they work through an ethical AI problem rather than simply asking them to report what they believe.
“First, we want to make sure the concepts are reasonable and validated by experts,” Tang says. “Secondly, we will use more nuanced analysis to make sure we can draw the most critical information from the relationships on the maps.”
The researchers ultimately hope their work leads to openly available resources that other faculty can use to incorporate ethical AI literacy into engineering and other STEM courses. Wang also sees opportunities to expand the approach into other engineering fields such as manufacturing and nuclear engineering.
The project is the first collaboration between Wang and Tang involving USC's Molinaroli College of Engineering and Computing and the College of Education. For both researchers, bringing their disciplines together is essential as AI becomes more embedded in engineering.
“It’s important to keep the engineering and education research parallel in this area,” Tang says. “We can have innovations in engineering, but at the same time, when we translate that innovation into preparing the future workforce, education research is needed.”
That balance is particularly important to Wang. While education can sometimes be a supporting component of engineering research, he wants it to remain at the center of this project.
“It’s developing an education tool with some novel research, but with the mindset of how we can transition it to the classroom and investigate ethical AI literacy,” Wang says. “I want to keep the education and ethical AI literacy equally as important.”