By Demian Hommel, CTL AI in Teaching and Learning Fellow in partnership with the AI Literacy Center
As part of the Intentional AI @ OSU series, I sat down with Jonathan Leong, an instructor in finance at the Oregon State University College of Business. In the high-stakes world of finance, Jonathan has found that generative AI is not just a tool for efficiency, but also a bridge to better student support in the asynchronous Ecampus environment.
Jonathan’s approach is defined by targeted automation: using AI to handle the grunt work and foundational questions so that both he and his students can focus on high-level financial analysis.
The challenge: The 24/7 support gap
In the asynchronous environment of Ecampus, students often work late at night or on weekends when an instructor isn’t immediately available. In a technical field like finance, hitting a wall on a Week 3 module can stall a student’s entire progress. Additionally, creating the complex, data-rich scenarios required for portfolio management — tasks that often take a week of manual labor — presented a significant hurdle to providing fresh, relevant content every term.
The innovation: Modular chatbots and 50-page simulations
Jonathan addressed these challenges by building specific, restricted AI assistants and leveraging high-level scenario generation:
- The modular 24/7 TA: Jonathan creates custom chatbots “fed” only with specific course materials, such as a single week’s module or a discrete set of problems. By restricting the AI’s context, he minimizes hallucinations and provides students with a high-accuracy assistant that can guide them through calculations and “easy questions” when he is offline.
- High-stakes scenario generation: For his portfolio management class, Jonathan used Google Gemini to transform a three-page data set into a comprehensive, 50-page financial simulation. This allowed him to present students with complex, realistic scenarios that would have previously taken him a week to draft manually.
- Administrative scaffolding: Jonathan uses AI to generate initial slide deck outlines and to rewrite multiple-choice question variables each term. This ensures academic integrity across different sections without becoming an overwhelming administrative burden.
Reflection: The “expert-in-the-loop” requirement
Jonathan is quick to remind his students that AI is an “ingestion engine,” not a creator. He compares the technology to a friend who “lies to you 10% of the time.” While AI can handle calculus, it often struggles with the specific logic of spreadsheets, probability, and statistics unless the parameters are perfectly defined.
It’s about eliminating the grunt work. If I can have a tool that helps me conceptualize and present ideas faster, that’s a useful skill. But you still need the foundational knowledge to know when the output is complete garbage. — Jonathan Leong
Key advice for faculty
- Prioritize the “expert-in-the-loop”: Teach students that AI is an “ingestion engine” that requires a human verifier. Frame it as a “friend who lies 10% of the time.” If students don’t have the foundational skills to judge the output, they risk accepting financial models that are “complete garbage”.
- Reclaim the “instructional 20%”: Use generative AI to automate the administrative grunt work of teaching such as drafting slide deck outlines, rewriting multiple-choice question variables, or organizing resource lists. Automating these logistical burdens can shave 20–30% off instructional prep time, allowing you to focus on high-impact student engagement.
- Teach debugging over execution: In a professional world where AI handles low-level analyst tasks, shift your instruction toward logic and troubleshooting. For example, when using AI for coding in Python or building financial models, focus on teaching students how to identify and debug the AI’s inevitable errors rather than just generating a final answer.
- Counter AI fatigue with high-value scenarios: Students may become discouraged by AI hallucinations and stop using the tools altogether. Counter this fatigue by demonstrating high-level applications, like turning a three-page data set into a massive 50-page financial simulation, that show the technology’s power to handle complex scale beyond simple Q&A.
About the Author: Demian Hommel is a professor of geography and environmental science in the College of Earth, Ocean, and Atmospheric Sciences and is an AI in Teaching and Learning Fellow with the OSU Center for Teaching and Learning. When he isn’t exploring the societal and environmental impacts of AI, you can find him DJing under the alias Dr. Gonzo or trying to graft citrus trees in his greenhouse.
Top image generated with Microsoft Copilot.
Leave a Reply