U of A official: Artificial intelligence should help, not replace, workers

"At the university, our goal is to grow the success of our students, to grow our research, and have an impact across the country, and our team that’s doing that is constantly asked to do more,” David Ebert, who is also UA's chief data science officer, told the Arizona Daily Star in an interview. “We want to make them more effective and use their skills effectively and enable them to grow the university.”

The UA launched its “AI in a Domain” leadership cohort earlier this year, which currently includes 49 faculty and staff members from across the university in a three-tiered program. They are working to create what Ebert calls a map for how to “responsibly and effectively” use artificial intelligence across education, research and university operations.

The cohort was created to get faculty, staff and students in different disciplines engaged in conversations about how AI should be assimilated into university functions, as well as to look at what the UA has done in the past and what universities across the country are doing.

The roadmap, which will guide the UA for six months at a time, is a living document that will change every semester and will be opened up to the broader university community for feedback, Ebert said.

While the inaugural cohort includes only faculty and staff, Ebert said they’d like to bring in students and their “brilliant ideas” as the program develops in the spring of 2027 and the following year. He said his office already reaches out to student government associations and graduate student associations at UA to get input, and he wants students to have an “active opportunity” to be engaged in the program.

AI is constantly changing

Ebert said the UA is executing its cohort's plans in short segments because of the pace at which AI is changing. The work is born from the “needs, ideas and problems” that the university community sees, he said. 

“Which is why we hold two town halls a semester, we have all of these sessions on a weekly and biweekly basis for people across our campus,” he said. “We had 250 people apply across the three programs — (those) being our AI Campus Champions, our AI Fellows and our AI Associate Directors. And we selected 49 people into that program so that we get good representation across the campus.”

Out of the two town halls held at the UA each semester, one is a listening session, and the other is to update resources available across campus.

The weekly and biweekly AI information sessions bring in three people each session who spend an hour talking about how they’ve used artificial intelligence in a new way.

AI use in education  

Ebert said the UA has had guidelines on the usage of AI for over a year, and has a “three-phase approach” to the use of AI: Courses where there’s completely free use of AI as long as it’s documented; courses where AI is used in certain tasks or assignments with documentation; and others where AI isn’t used; and finally, courses where there is no use of AI at all.

“What we want to do is find the best practices, really look long-term on how we think the educational process needs to change, and from that, come up with a strategic approach of how we see the educational process in universities transforming over the next several years,” he said. “Because our goal is the success of every student, which really means determining, ‘How do we ensure that the students that we graduate have the skills that will last them for the next 30 to 50 years in their career?’”

He said there are many faculty members in data science, business, cognitive science, humanities and medicine who are actively engaging with AI to increase students’ learning and give them the skills they need.

An example provided by Ebert's office: Randi Weinstein, a senior lecturer in UA’s Department of Physiology, develops AI-simulated patients, physicians and clinical assistants to teach students about differential diagnosis and clinical reasoning. She creates interactive websites for students to study “complex biosynthetic and developmental pathways,” which are multi-step biological processes that organisms use to build essential modules and grow from a single cell into a mature organism.

Other ways in which Weinstein uses AI include using it to brainstorm and create case studies according to the learning objectives she has for students; teaching AI skills to her honors and graduate students so they can create their own interactive websites; and brainstorming ideas for in-class and out-of-class activities.

Ebert said his office is also collaborating with UA Provost Patricia Prelock’s office to create an education task force. It will host a series of conversations with groups across the UA to discuss ideas, problems and best practices, so they can come up with guiding principles for how to move forward with AI usage in education.

“It’s key from all employers that I’ve talked to that there’s fundamental knowledge (about AI) that people need to have in their career,” Ebert said. “Even if an AI tool will help them become more effective in automating certain tasks, they need to know what are the right values, what is the series of operations, and (how) to determine whether or not something they’re doing is correct.”

Matt Stoner, a UITS senior information systems manager and adjunct professor in the College of Engineering, has worked in the teaching and use of AI in the Technical Communication general education course he teaches, Ebert’s office said, giving another example. The main focus of his teaching is to make students understand how to use critical reasoning and judgment skills to navigate the use of AI.

As part of his course, he’s teaching students to be fluent in AI, so they’re prepared for the workforce and is bringing AI into course activities to help students improve their public speaking, writing and teamwork skills. 

Ebert said that through the “AI in a Domain” leadership cohort, UA is coming up with guidelines on who gets to make decisions about how AI should be incorporated into a given class, course or degree. 

“You can have two faculty members who come at the same topic from different perspectives and backgrounds, that see the role of AI differently,” Ebert said. "We’re trying to set a baseline and guidelines for people to decide those principles effectively, and they’ll be reflected in the requirements for the degree and every faculty. And that will be done individually by department and by college.”

AI use in research  

As for the incorporation of AI into research, Ebert said, “We have our AI innovation pilots, where people across the university can apply, whether they’re looking at the operations of the university, doing something new and interesting in their classroom, or trying to take and introduce machine learning or large language models or some technique into their research to help them become more effective.

"We evaluate those, and we have a staff of 25 people who can work with them for six months to help them get that prototype and figure out how to deploy it in the systems they’re doing.”

Ebert said that while bringing in AI exploration, it’s important to also consider privacy and data security.

‘A cultural change’

Ebert said he doesn’t think AI can replace the academic instruction that faculty members and universities offer students. Faculty instruction is key, and there’s much information that a faculty member can provide that an AI tool cannot, he said. AI is just another learning aid, like having a tutor in your class or having an introductory lab assistant that helps with research, he said.

“It’s important for everyone to understand what this is and how to use it, and not to rely on it,” he said, “but to use it as just another resource, as they use searching for information on the web or reading multiple opinions on a given topic to help them understand something in a contested issue.”

Addressing the inappropriate or negative uses of AI at universities, including plagiarism, Ebert said plagiarism has been an issue in society for a very long time. The job of educators is to teach students to look at AI and other tools not in terms of replacing the work they need to do to learn, but to use it as a resource to make that same learning more effective, he said.

“I’m someone who doesn’t focus on the negative side of things, and I think this is something that happens with every technology,” Ebert said. He sees the AI evolution “as a cultural change … and as long as we are able to distill that cultural change, I don’t think we have issues to worry about.”