Students can respond to case study prompts created by professors with a generative AI model called STRATPATH.
Nicole Coomber has taught consulting and experiential learning courses at the University of Maryland’s Smith School of Business for years, assigning graduate students take-home case studies that mimic consulting interviews.
But, like many professors in the ChatGPT era, Coomber found that homework no longer posed problems for his classes, because students simply typed questions into large language models and submitted whatever the generative AI model spit out.
“I found that students could virtually take my assignment, plug it into the AI, and get a perfect answer without having to face some of the challenges that we know are part of learning,” she said.
Case studies are an essential part of the interview process for many business students. So it’s important for Coomber to make sure they participate in the exercises and don’t bypass critical thinking. But rather than creating a new low-tech mission, Coomber teamed up with a group of master’s level students to create an AI tool to act as a case investigator.
The result is STRATP ROADa generative AI tool that offers faculty-created case studies to evaluate and provide real-time feedback to business students. The tool connects student learning to real-world scenarios and provides career readiness skills, preparing students for interviews after graduation.
How it works: STRATPATH was developed by six recent graduates of UMD’s business school: Deep Dalsaniya, Anna Huertazuela, Aditya Kamath, Aromal Nair, Krishang Parakh and Venkatesh Shirbhate. The team first came together to compete in a case competition for MBA students in 2024, then returned to the university after graduation to launch STRATPATH, using funds allocated by the dean.
“It was kind of like, ‘Hey, these are really talented students, the job market is really tough, they could use a soft landing to follow their job search,’” Coomber said. “It’s become something much more; we’ve built something truly incredible.”
Students can chat with STRATPATH or respond via audio to the faculty-developed case study.
To set up the tool, professors provide the story of the case study they want the student to answer, a rubric or feedback form, and some examples of ideal answers, Dalsaniya said. Depending on input, STRATPATH facilitates prompts via audio or text, engaging the student in conversation.
“Students prepare themselves to think spontaneously and develop their critical thinking skills,” Dalsaniya said.
STRATPATH relies on an extensive language model with additional limits set by developers to reduce the chances that the AI will hallucinate, accept incorrect information, or provide overly complimentary feedback. He also studies students’ answers to ensure they are not cheating by using outside sources.
“He doesn’t say, ‘Deeply, you’re so smart, it’s true!'” Coomber said. “It’s like, ‘How did you get here?’ “So even if students type into ChatGPT and then put that answer into our platform, our platform will say, “How did you get to this?” »
The platform also doesn’t allow copying and pasting answers, so if a student abstains from ChatGPT while answering STRATPATH, they must at least transcribe the answers (and at a reasonably human rate of words per minute), which will hopefully produce learning to some extent, Dalsaniya said.
“Our main goal is to find out whether or not their critical thinking skills increase, even if they cheat,” he said.
The impact: STRATPATH provides instant grading and personalized feedback in real-time, saving teachers time and helping students adapt faster.
Previously, Coomber would take hours to review student assignments, which could interfere with learning due to the long delay between assignment and feedback. She can now spend more time on face-to-face learning or office hours.
Anecdotal feedback from students so far indicates that they feel better prepared to approach interviews and have appreciated the tool’s assessments, which identify both areas in which they excel and areas in which they could improve.
What’s next: Coomber and his team are looking to identify other campus stakeholders who might have a use case for STRATPATH. One option is to work alongside the career center to provide behavioral interview prompts. Many interviewers ask candidates to use the STAR (situation, task, action, and result) method to answer questions and use it to assess talent, and STRATPATH could be a forum for students to practice these questions.
Dalsaniya and the development team are also exploring ways to feed STRATPATH additional resources from faculty to provide a richer assessment of student responses to case studies.
“Case-based learning has no right answer: all answers can be right,” Dalsaniya said. “What we’re trying to focus on is how we can embed all of the professor’s lecture materials, including their slides, their video lectures, into the comments so that the student can see the comments and reference those slide numbers or those chapters or those video transcripts. »
The team is also seeking additional funding sources to scale the tool and potentially license the tool to outside groups.