91��ý

Bobby Hodgkinson is exploring the pluses – and minuses – of generative AI in academia.

An associate teaching professor in the Ann and H.J. Smead Department of Aerospace Engineering Sciences at the University of Colorado Boulder, Hodgkinson is working to steer positive AI adoption in education.

“People are starting to feel the effects of AI in business,” Hodgkinson said. “How can we train people to work alongside these tools and position themselves? We owe it to our students to explore what these tools can do well.”

Those efforts are taking multiple forms. Hodgkinson is working to roll out AI tutors for assignments, grading assistance, and data analysis. As AI is still in a state of rapid evolution, some initiatives have gone better than others.

“Accuracy isn’t to the level that we need AI to be, but it’s about 80% and this is the worst it’s going to be. I’ve said to students, ‘I want to expose you to this because you’re going to be entering this new society and I want you to know what it’s good at and where it struggles’,” Hodgkinson said.

He sees particular potential for AI to improve personalization of learning.

“In class, the intent of a test is for me to understand where you are in the learning process. If you give me an incorrect answer I know something is up, but I don’t know what it is. If I can add a box for you to explain your thinking in your own words, I can get a much better idea of where you are, but that is very hard to administer at scale to 100-plus students. An AI can analyze those responses and I can start directing my intervention,” Hodgkinson said.

He is already putting it into practice with student lab reports.

“Instead of a grader having a 25 page lab report they have to review 50 times for 50 different groups, the AI tells the grader this is where you should focus your attention,” he said.

He is currently finishing up work on a paper, to be published by the American Society for Engineering Education, evaluating a class project that incorporated an AI component. Hodgkinson has also led seminars for other aerospace faculty on applications of AI in education.

A major concern with AI systems has been student cheating, but Hodgkinson has an unusually positive attitude.

“If we’re just asking students to do something a machine today can do, I’m cheating them out of an experience. AI is amazing at writing computer code. The expectation of an entry level engineer to only write a few hundred lines of code a week are gone. But AI is not good at creating the architecture for these applications – what does the client want and how do we turn that into tangible tasks,” Hodgkinson said.

As AI tools have advanced, Hodgkinson has found them to be tremendously beneficial in his professional and personal life – helping to write emails and summarize complex concepts, improving his efficiency and allowing exploration of ideas he previously did not have the bandwidth to tackle.

“While I’m walking my dog I’ll be chatting with ChatGPT through my headphones, reasoning through an idea I have,” he said. “I could have that discussion with another person, but when you talk to a person, you’re always thinking about how you will be perceived. Interacting with a machine removes that.”

Hodgkinson is an active member of the , a community hoping to advance uses of “AI for good.” It has helped him connect with other university faculty and K-12 educators who are also experimenting with AI.

“The goal is education focused more on the individual student,” he said. “Where are you today, where do you need to be tomorrow, and how do I help get you there. We didn’t have the resources to do that before; there’s not enough hours in the day. Now if I can build a tool, I can do something about it.”��