Teacher Bárbara Cleto is leading a project with her 11th-grade Vocational Education students for the Digital Systems course – for the short training unit on Algorithms and Introduction to Programming using Arduino. During 4 lessons, she invites students to create and validate assessment tests using the generative AI tool ChatGPT.
First part
The teacher began the lesson by presenting the activity to the class: create an assessment test based on the content studied, using ChatGPT.
The teacher began by demonstrating how to use ChatGPT and how to generate different types of tests:
- Defined examples of structure (multiple choice, true/false, short answer).
- Demonstrated how to ask the chatbot to correct a test.
- Compared the different results obtained.
- Showed examples of more and less effective prompts.
Second part
Next, each student accessed ChatGPT to create their own test:
- Define the structure and content to be included.
- Write complete and clear prompts.
- Interact with the tool to generate the test.
- Adjust the questions to ensure they accurately reflected the content studied.
Throughout the activity, the teacher monitored each student individually, helping them improve the clarity and relevance of the instructions given to the generative AI.
In this lesson, students solved the tests they had co-created with ChatGPT in the previous lesson.
During validation they were required to:
- Check whether the questions were aligned with the content taught and with the lesson notes.
- Reword the questions when necessary.
- Confirm the correctness of the answers generated by the AI.
- Use the Flowgorithm and Tinkercad tools to review specific technical content.
- Compare the AI’s answers with their own notes and corrected any errors identified.
In this session, each student exchanged their test – which they co-created with ChatGPT – with another classmate’s test:
- They completed the test they received.
- They corrected the answers with the support of ChatGPT.
- They compared their answers with the corrections generated by the AI.
- Together with their classmate, they analyzed the consistency of the questions and answers, identifying any errors, ambiguities, or inconsistencies.
This peer feedback dynamic promoted critical analysis and constructive discussion about the use of AI as an assessment support tool.
In this session, the teacher led a group discussion, encouraging students to share their experiences, reflections, and conclusions. The questions addressed were:
- How were the instructions given to ChatGPT?
- What adjustments were necessary during the process?
- What errors or limitations arose in the AI’s responses?
- How was the interaction with the AI?
- What lessons and challenges did this activity bring?
The final dialogue allowed for the consolidation of learning and the development of critical awareness about the ethical, critical, and pedagogical use of AI in supporting teaching and learning.
Conclusion
Throughout the 4 lessons, students went through all stages of the AI assessment cycle – create, validate, exchange, and reflect – reinforcing their understanding of the content and developing skills in autonomy, digital literacy, and critical thinking.
Teachers’ shared reflections
The students were surprised to see when ChatGPT gave incorrect or incomplete answers.
They (students) enjoyed creating a test with the help of AI, even though they sometimes had difficulty knowing “how to ask” for what they wanted. (…) One of the biggest difficulties for students is knowing how to create prompts and then “continue” the conversation.
They (students) mentioned that it was different to create a test and have to solve it, even when they already had the AI correction, because sometimes it was necessary to make adjustments to that correction and “teach” ChatGPT.
It was a different way for students to check whether they had consolidated the content, and it was found that students were more engaged.
The use of AI proved useful, but it was found that some students did not correctly validate the answers given by chatGPT.
