Studying with Gemini

generating quiz questions

Posted by on June 20, 2025 · 2 mins read

My sons have been studying for their final exams this past month as school is wrapping up for the summer. While their teachers have sent home some practice questions, they felt they needed additional practice on a handful of specific topics. Since, I haven’t been in school for many decades, my personal ability to help with some of their math problems was limited to say the least. :)

Summarizing a Topic

Instead, my younger son stumbled across a help feature of Google Gemini (and presumably others like ChatGPT), when asking for a summary on a challenging part of his upcoming Math exam:

As expected, this provided a good breakdown of the geometry subject he wanted clarification on, beyond the notes that his teacher had provided.

Generating a Quiz

Next, he simply prompted for “practice questions” based on the earlier summary.

Surprisingly, this generated an entire mini-quiz of ten questions that appeared in a separate dialog area:

He was then able to work through the questions and see the result. In particular, he got one wrong and Gemini then offered to further discuss the theory behind the correct answer.

Not Perfect

Since the topic was geometry, he asked for an image to illustrate Thale’s Theorem, covering the angles between 3 arbitrary points on the diameter of circle. Here is the AI-generated summary and image of the theorem:

Do you notice that the generated image failed to place the corners of the triangle on the circle? This actually confused him, so he issued a plain Google Search query and got images created by others which did show the correct point positions.

Conclusions

Overall, Gemini was a really helpful tool for my son during his exam study session. The mistake in generated image was also a good opportunity for him to realize that he needs to be able to judge the accuracy of all AI generated content. I am trying to support my son’s in their responsible use of AI as a study aid, a tool in essence. However, they still need to understand the topics themselves and make good judgements on AI’s output, always with a sceptical eye. Hopefully, this will help them avoid the current “cheater’s dilemma” facing students at all levels these days.