Collaborative Projects with ChatGPT

AI as a Collaboration Catalyst

Collaborative projects are the training ground for transversal skills: communicating, organizing, and solving problems together. You might worry that AI has no place here, since the goal is precisely to get humans working together. Yet, positioned correctly, ChatGPT can act as a facilitator that strengthens teamwork rather than replacing it. The key is defining that role precisely; otherwise, the tool absorbs the very work the project was meant to generate among students.

These projects, where students solicit the tool themselves, fall within the framework established earlier in the course: in class starting from 9th grade (4e), and with supervised autonomy in high school. One device per group, used without an account and without any personal data, is sufficient. Before 9th grade, the teacher queries the tool on behalf of the group, projecting it on the board.

What AI Brings to the Group

During brainstorming, AI generates a volume of ideas the group might not produce alone, but the group must sort, refine, and select them. This deliberation is itself an exercise in collaboration. During the research phase, it accelerates the collection of basic information—which must be verified—freeing up time for what only students can do: field investigations, interviews, and observation.

When ideas accumulate in disarray, AI helps structure them into an action plan. It can even act as a mediator: when two subgroups dig in their heels on incompatible options, asking it to formulate compromises unblocks the situation without any student feeling they have “lost.” At the deliverable stage, it assists with formatting—assists, and does not write. This is a distinction you must not compromise on.

Four Project Formats That Work

If you are just starting out, pick just one format and adapt it to your subject and grade level; the others will come once habits are established.

The class newspaper is the most accessible. Students create a newspaper around a theme. ChatGPT helps structure sections, identify claims in articles that need verification, improve writing without rewriting it, and suggest angles no one had considered. The non-negotiable rule comes down to one sentence: every student remains the author of their article. AI occupies the post of virtual editor-in-chief, the one who asks the right questions—what is your angle, who is your source—without ever holding the pen.

The structured debate follows a five-phase sequence where AI’s role is strictly delimited:

1. ChatGPT presents the topic, and the class verifies that the presentation is balanced
2. Each team asks it for arguments supporting their position
3. Students select, verify, and rephrase the arguments
4. The debate takes place without ChatGPT
5. ChatGPT helps draft the collective summary, which the class validates

Note the fourth phase: the central moment, where students argue in real time and absorb contradiction, is deliberately AI-free. Students arrive better armed, but they are on their own, and that is what gives the exercise its value.

The interdisciplinary inquiry mobilizes several subjects around a local issue, such as the water quality of a nearby river or the history of a neighborhood. Its value lies in the explicit distribution of roles at each stage:

Finally, the educational hackathon concentrates energy into a single day: teams solve a challenge with ChatGPT’s help—designing an awareness campaign, imagining a service that improves neighborhood life, or writing a guide to their city in a foreign language. The short format forces efficient use of the tool rather than getting lost in it.

Organizing Teamwork with AI

Without explicit organization, a group placed in front of ChatGPT quickly reduces to one scribe and three spectators. The solution is to distribute named roles, announced before the session. The Researcher queries the tool. The Verifier checks the reliability of responses against sources, a mission that teaches methodical doubt far better than a lecture on model errors. The Writer produces the final content, by hand. The Coordinator uses the tool to organize the group’s schedule. Rotate these roles from one session to the next.

Also prepare a logbook template, kept by each group, where they record requests made to the tool, responses obtained, what was kept, modified, or rejected, and especially why. This logbook is an instrument of metacognition: by explaining why they discarded a response, students formalize their critical thinking. It also provides you with the best window into the group’s actual process.

Evaluating Collaborative Projects

Evaluation focuses on the process as much as the result. The quality of collaboration is observed during the project: who participates, how decisions are made, who stays silent. The relevance of AI usage is visible in the logbook, where a group that rejects a response with argumentation demonstrates real competence. Added to this are the originality and quality of the deliverable, followed by the group’s ability to present and defend their work orally, the moment when you see who understood what was produced.

These criteria are valid only if they are set and announced before launch, along with what is permitted: the Ministry considers the use of AI for an assignment without explicit authorization or personal work of appropriation to be fraud.

Your Turn

Launch a first project.

1. Choose a format—newspaper, debate, inquiry, or hackathon—and a grade level starting from 9th grade
2. Define the four roles and their rotation
3. Prepare the logbook template, with no personal data
4. Write the evaluation criteria and what is permitted, then announce them
5. Provide one device per group, tested without an account on the school network

Official sources: Getting started with ChatGPT as an educator · Searching the web with ChatGPT (OpenAI Help Center) · Artificial intelligence and its uses in education (éduscol) · Publication of the framework for the use of artificial intelligence in education (Ministry) · Teacher: how to use an AI system in the course of your duties? (CNIL, France’s data protection authority)

Key Takeaways

• AI facilitates collaboration but does not replace interaction between students
• Every group member has a defined role, including that of Verifier
• The logbook develops metacognition and reveals the actual process
• Evaluation covers both process and result, with criteria announced in advance
• These projects require a framework: students in class from 9th grade onward, one device per group without an account