Human or Machine? Students exchange texts, make assumptions about their origins, and reflect on credibility and how AI shapes academic writing.
- Group activity
- Human-AI-comparison task
- In class or online
- GenAI-assistance in academic writing
- All disciplines
- Basic
- 45-60 min
- approx. 30 / small groups of 3-5
- GenAI tools
- Flexible classroom setting
Short description
In this activity, students each produce a short written response to an academic question, either independently or with the help of a generative AI tool. They then exchange responses within small groups and attempt to identify which texts were AI-assisted and which were purely human-written. After the guessing game, the group engages in a discussion about which features made a response seem credible or gave away AI involvement. Through comparison, discussion, and reflection, students develop a more critical understanding of AI‑assisted writing and its implications for academic practice and integrity.
Competence domain of the Didactic Framework: Critical Engagement
By the end of this activity, students can…
- compare AI-generated and human-written responses to an academic question, identifying differences in tone, style, and content quality. (FLAIR Didactic Framework: LO8)
- develop evaluation strategies for human and AI written texts by assessing their credibility and quality using criteria such as factual accuracy, coherence, and use of references. (FLAIR Didactic Framework: LO6)
- identify and judge whether texts were likely AI-assisted or human-produced based on features such as overly generic phrasing or lack of personal voice.
- reflect on their own use of AI in writing and articulate how AI assistance and over-reliance can affect the writing process, academic integrity, and learning outcomes. (FLAIR Didactic Framework: LO9)
Instructions
Prepare an academic question or prompt related to recent course content that allows for a focused but substantive response (e.g. a short essay paragraph or a brief explanation). Divide the class into small groups of 3-5 students and evenly assign group members to one of three writing conditions: written without AI support, written in collaboration with AI or fully AI‑generated. This ensures that all categories are represented within each group and enables meaningful comparison in later stages. Instruct students not to reveal their assigned condition to peers during the individual writing phase.
Students write their response according to the assigned condition. Clear guidelines for length and time (e.g. 150-200 words; 20-25 minutes) help balance depth and feasibility.
To support quality and comparability, this writing phase can take place at the end of one session. Collect the texts and compile them into a shared, anonymized format for the next steps. If some students need additional time, they may complete their response as homework.
Ask students to exchange their responses within their group. If you have collected the texts in advance, provide the group with a compiled and anonymized set of responses. Otherwise, students can exchange texts directly by passing papers around, swapping devices, or posting all responses on a shared platform in an online setting.
Ensure that each response is labelled without indicating the author or writing condition. Ask every group member to read all responses from their group before moving on to the guessing phase.
For each response, the group discusses and guesses whether the text was written without AI support, with AI assistance or fully generated by AI. Encourage students to justify their guesses by pointing to specific clues in the text, such as tone, level of specificity, structure or presence/absence of a personal voice.
Instruct authors to remain silent while their own response is being discussed to avoid influencing the group’s judgment. Once the group has agreed on a guess, ask the author to reveal how the text was produced.
Once all texts are revealed, the group reflects on the experience. It is recommended to provide guiding questions: What features made texts seem human-written, AI-assisted or fully AI-generated? Were there any surprises? Which texts were seen as most credible or high-quality, and why?
Also encourage students to consider how knowing a text is AI-generated (or not) influences their trust in it, and what this implies for evaluating academic writing and AI use more generally.
Invite groups to share key insights, surprises, or points of disagreement, and synthesize common patterns observed across groups. Emphasize that distinguishing AI-assisted from human writing is not always straightforward and tie the activity back to course objectives (e.g. the importance of critical evaluation of sources, or ethical use of AI in academic work).
Assessment
This activity is primarily intended as a formative learning experience in which students learn through the process rather than aiming for a grade. If assessment is required, it is recommended to focus on students’ engagement in group discussion and on short individual reflective tasks addressing what they learned about evaluating texts, AI assistance, and academic integrity. The emphasis should remain on developing critical evaluation and awareness.
If connected to course content, written responses may additionally be reviewed through instructor or peer feedback, focusing on relevance, accuracy, and alignment with the task, regardless of AI use.
Possible challenges
- Students may be tempted to rely on AI‑detection software rather than close reading and discussion.
- Texts might vary in length and discussions might be rushed due to time constraints.
- Difficulty distinguishing AI-assisted from human writing may lead students to assume that AI-generated text is acceptable in all contexts.
- Group discussions may focus repeatedly on a small set of familiar indicators of AI use, limiting the diversity and depth of analysis.
How to adress them
- Emphasize human judgement and discussion. If time allows, briefly test one or two AI‑detection tools together. Use the results to discuss limitations of such tools.
- Set clear guidelines for text length and timing to keep the activity manageable.
- Clarify when independent writing remains important and why. Use brief examples of AI limitations or failures (e.g. hallucinations or inaccuracies) to reinforce responsible and context‑appropriate AI use.
- Use a simple guiding framework to encourage students to consider a wider range of features (e.g. language use, structure, evidence, specificity, coherence, or presence of inaccuracies), without presenting it as a definitive checklist.
- If needed, the instructor may provide an additional AI-generated response to ensure a balanced range of writing conditions within each group and put focus on certain AI characteristics
- To encourage participation and individual judgement during the guessing phase, students can first write down their individual guesses before discussing them as a group
Waltzer, T., Pilegard, C., & Heyman, G. (2024). “Can you spot the bot? Identifying AI-generated writing in college essays.” International Journal for Educational Integrity, 20(1), Article 11.
Using this resource
This resource is licensed under Creative Commons BY-NC-SA 4.0 license. Suggested citation: Flair Collaboration. (2025). FLAIR Toolkit. Teaching GenAI Competencies.
