NEWS · MODELS · #1549
Malaysian lecturer swaps traditional coding tests for Gemini‑assisted verbal assessments
A Chief Digital Officer and lecturer at Universiti Putra Malaysia replaced conventional coding exams for a data mining course with conversational assessments where students use Google Gemini in Colab to generate code and must verbally justify model choices; the instructor uses Gemini Canvas and Gemini Notebook to evaluate work at scale. The approach emphasizes students' judgment and troubleshooting over raw syntax skills.
KEY POINTS
- A Chief Digital Officer and lecturer at Universiti Putra Malaysia replaced conventional coding exams for a data mining course with conversational assessments where students use Google Gemini in Colab to generate code and must verbally justify model choices; the instructor uses Gemini Canvas and Gemini Notebook to evaluate work at scale.
- The approach emphasizes students' judgment and troubleshooting over raw syntax skills.
- This example shows a practical shift in assessment toward evaluating human judgment in human–AI collaboration rather than manual coding ability, with implications for how computing education adapts to powerful generative models.
WHY IT MATTERS
This example shows a practical shift in assessment toward evaluating human judgment in human–AI collaboration rather than manual coding ability, with implications for how computing education adapts to powerful generative models.