AI Agents in Education: What’s Working and What’s Missing

Source Domain: www.govtech.com

Across universities, agentic artificial intelligence is still in a trial phase. Pilots in advising and tutoring and integrations into learning management systems (LMS) represent a sector dipping its toes in a relatively new technology.

The term “AI agent” has only recently entered mainstream education conversations, gaining traction around 2023-2024. According to Nate Ober, senior ed-tech and AI/machine learning leader at Amazon Web Services (AWS), it refers to a system that can use tools, like databases or LMSs, to plan a sequence of steps, take action, observe results and adjust, operating in a loop until a desired goal is completed.

Early pilots struggled with workflows that included more than a few steps, making implementation look similar to that of generative AI: retrieval-based Q&A and content generation, Ober said in an email to the Center for Digital Education. Now, agents’ capabilities have evolved to allow them to work independently for hours.

“We are in the earliest days,” said Nicole Engelbert, vice president of product strategy for student systems at Oracle. “Take a side eye on what anyone is saying about what’s happening in a pervasive way.”

EASY WINS: ADMINISTRATIVE SYSTEMS

Experts say the strongest easy wins for agentic AI are administrative, where the work is high-volume and repetitive.The Illinois Institute of Technology, for example, automated transcript processing in 2023. The automated system handles intake, international grade conversion and customer resource management integration, reducing processing times from roughly a month to a single day.

Similarly, at Highline College in Washington state, a financial aid status tracker introduced in 2023 reduced emails, phone calls and in-person visits about application status by 75 percent.

These systems are also appearing inside LMSs. For example, Instructure’s Canvas system now has an agentic tool that follows…

Source