3 minutes

Explain It: Jonathan Kalodimos' new AI-assisted oral evaluation tool

Jonathan Kalodimos was watching two trends collide: Students were increasingly using AI to generate written coursework, and faculty were losing confidence that traditional assignments reveal what students actually know.

His response was Explain It, an AI-assisted evaluation platform designed to work in partnership with faculty, built around a format AI can't fake: oral presentations.

“I thought, 'We have all of these tools available and I can code,'” he said. “I wanted to ensure that I was robustly extracting the evidence of learning in parallel with what I call our fallible senses.”

So Kalodimos, an associate professor of finance and Harley & Brigitte Smith Fellow in the College of Business, developed Explain It, an AI assessment tool designed to be a “partner” for faculty.

Students record oral presentations explaining their analysis and submit them along with their slides. The platform, developed in partnership with Metrum AI, transcribes students' speech and analyzes their slide content, including text, figures, tables and charts, then evaluates both against the assignment rubric to identify specific evidence of learning. After Kalodimos grades the presentations independently, he can compare his work to Explain It’s to make sure he’s been as thorough and fair as possible.

The tool, Kalodimos said, goes beyond surface-level topic coverage. It assesses the depth of a student's reasoning, distinguishing between a student who recites a definition and one who can apply the concept to a new situation.

“Teaching assistants and professors are grading traditionally with a detailed rubric, and there’s a parallel track where they copy and paste feedback where appropriate. I’m using my professional expertise to determine what’s valid and what might not make sense," Kalodimos said. "My goal is to provide meaningful feedback to students. We can use the current technologies to our advantage to do better.”

Explain It, Kalodimos said, is not designed to produce a grade. Rather, it is designed to evaluate learning. Explain It generates a draft evaluation citing specific evidence. The instructor reviews, modifies or overrides the AI's assessment.

“The professor is always the decision-maker. Faculty retain full authority over every grading decision," Kalodimos said. “This is not acting as a substitute, and it’s going to be interesting how the human grading element compares to the tool assessing where they are in their learning process."

Kalodimos piloted Explain It in his own classes and received positive feedback from his students. He then presented Explain It at Oregon State’s AI Week, which brings the OSU community together in partnership with industry, in April 2025.

The presentation intrigued campus partners, and Explain It’s pilot grew organically. Currently, faculty are using Explain It in the Colleges of Business; Engineering; Earth, Oceanic and Atmospheric Sciences; and Health. A full campus-wide deployment is expected in Spring term.

“We have the opportunity to shape this, and as we learn more about it, we can keep improving it,” Kalodimos said. “We can all bring our expertise in to produce the best thing for our students.”

For Kalodimos, Explain It is a tool that can be crucial for educators at a time when AI technology is changing rapidly while adoption is also rising. His aim is to help faculty teach better.

“Right now, we’re in a ruckus trying to figure out how AI is going to affect higher education,” he said. “But once we figure out how to go forward, I think the learning experience for students is going to be more streamlined. We’re going to be able to accelerate their learning in some areas and slow down their learning in others so that they leave here with even more knowledge.” 

The work has drawn interest from industry partners including Dell Technologies and NVIDIA. Kalodimos is also leading a research effort to publish findings on AI-compatible assessment design, positioning Oregon State at the forefront of how universities adapt to the AI era.