A Generative AI Tool to Foster and Assess Authentic Learning: A Case Study in Teaching SQL
Raja Sooriamurthi, Xiaoying Tu, Allison E. Connell Pensky · 2025
Authentic learning refers to a student's metacognition about what they have learned and not learned. Building students' authentic learning is a primary goal of instructors. Educators also want to be able to assess students' learning. One way to do this is to have a one-on-one conversation with students in which the instructor probes their understanding with a series of questions that require the students to explain why they did what they did, possible alternative approaches, and the implications of their learning. Scaling this type of high-impact 1-1, human-led dialogues with larger classes is a challenge. The present study built a custom generative AI tool and tested its ability to have such a conversation with students. In this paper we report on an experiment which aims to compare students' course performance and attitudes between conditions of instructor-led and AI-led dialogues. We found that regardless of condition, students significantly grew in their self efficacy from the beginning of the semester to the end of each of the two feedback sessions. Additionally, students receiving dialoguing with the AI were significantly less nervous than students dialoguing with the instructor with no differences in other attitude measures such as how deeply they believed they understood their assignments. As such, the intelligent assessor generative AI tool offers the potential to provide scalable feedback to students regardless of class size, likely with little to no detriment to students' growing confidence in their skills.