I replaced traditional coding tests with real conversations about how my students solve problems using AI.
A professor replaced coding tests with AI-assisted problem-solving conversations to assess students’ judgment and adaptability in data mining.
The instructor shifted from traditional coding exams to AI-integrated assessments, using Google’s Gemini in Colab as a collaborative tool. Students receive AI-generated code suggestions but must critically evaluate, modify, and justify their choices to demonstrate understanding. This approach emphasizes human decision-making over rote technical execution in data mining tasks.
Verbal interviews replace written tests, requiring students to explain their reasoning, troubleshoot errors, and identify flaws in AI recommendations. The process highlights whether students can effectively guide AI rather than merely follow its output. The professor frames this as a shift from dependency on technology to intentional skill development.
Behind the scenes, tools like Gemini Canvas and Gemini Notebook assist in evaluating student work at scale. These platforms help the instructor assess reasoning and problem-solving without increasing administrative burden. The focus remains on cultivating human judgment in AI-assisted environments.
The professor balances this teaching method with their role as Chief Digital Officer, advocating for AI as a partner rather than a replacement for learning. The goal is to build lasting human capabilities while leveraging AI to enhance educational outcomes in data mining.