Students break down research projects down into its constituent tasks and systematically decide how tasks should be delegated between human and AI.
- Team-based workshop
- Gallery walk
- In class or online
- GenAI tool testing for human-AI task delegation
- All disciplines (best suited for project-based courses)
- Advanced
- 90 min / 2 sessions
- Flexible / small groups of 3-4
- GenAI tool
- Flexible classroom setting
- Human-AI collaboration framework
Short description
In this hands‑on, team‑based workshop students work on a short research project and systematically break it down into its constituent tasks. For each task they decide whether it should be carried out by a human only, supported by AI or fully delegated to an AI tool. By testing AI tools (e.g. large language models, scholarly search engines, or code assistants) in real time, students explore the affordances and limitations of generative AI. They justify their delegation decisions using explicit criteria, and reflect on how trust, ethics and creativity shape human‑AI collaboration.
Competence domain of the Didactic Framework: Collaborative Intelligence (Humans & AI)
By the end of this activity, students can…
- decompose a research project into discrete, meaningful subtasks and evaluate these tasks using explicit criteria (e.g. complexity, creativity, ethical or data sensitivity, AI maturity).
- make transparent and well‑justified delegation decisions between human‑only work, AI‑assisted work, and full AI delegation (FLAIR Didactic Framework: LO21).
- document and communicate their decisions in a structured Task‑Delegation Matrix, including evidence from AI testing and consideration of potential risks.
- reflect on how personal trust, ethical aspects and academic integrity norms shape human-AI collaboration and apply these insights to future projects or contexts.
Instructions
Pre‑class preparation (~ 45 min – asynchronous):
- Construct and share an agreed Human-AI Collaboration framework outlining possible decision criteria for task delegation (e.g. complexity, creativity, ethical or data sensitivity, need for human judgment, reliability of AI outputs, or risks related to hallucinations and bias).
- Select a suitable project context for the activity: either provide pre-defined sample projects, let students choose a project relevant to them, or use a project that is central to your course.
- Ask students to individually list every task they think the project entails (max 15 items) in a shared worksheet (e.g. a Google Sheet, using individual tabs).
- Provide a short demonstration (e.g. a 5‑minute video) showing how to run a quick query in the AI tool chosen for the task.
In‑class activity (~ 45 min – synchronous)
At the beginning of the session, remind students of the learning outcomes and introduce the task‑delegation categories using a clear colour key (e.g. human only = blue, AI-assisted = yellow, fully AI-delegated = red). Clarify that the goal is not to optimize for speed, but to make reasoned and transparent decisions about collaboration with AI.
Assign students to groups of three to four, either randomly or using a pre‑assigned roster. Ask groups to move to their tables or breakout rooms (in an online setting) and ensure that they have shared access to the collaborative worksheet.
In their groups, students first merge the individual task lists into a single shared master list. For each task, they then apply the decision framework to agree on an appropriate collaboration mode (human‑only, AI‑assisted, or full AI delegation). Where AI use is considered, students test the task with a brief live query and note the outcome. Based on this process, groups complete the Task‑Delegation Matrix by documenting the chosen collaboration mode, at least two justification criteria, evidence from the AI test, and a short note on potential risks. During this phase, circulate among groups to prompt them to stay focused and to monitor the quality and appropriateness of AI outputs.
Invite students to review other groups’ Task‑Delegation Matrices, either by viewing them on screen or in printed form. Set a clear time limit for the gallery walk and ask students to focus on understanding the delegation decisions made by their peers. As they review the matrix, students add brief sticky note feedback comments, for example highlighting interesting choices, noting potential risks, or suggesting alternative uses of AI. The goal of the gallery walk is to encourage comparison across groups and to surface different approaches to human-AI task delegation.
Bring the class back together and facilitate a short plenary discussion. Encourage students to share patterns they noticed across the different Task‑Delegation Matrices, such as tasks that were frequently delegated to AI or consistently kept human‑led. Use this discussion to surface underlying assumptions, points of disagreement, and potential ethical or practical red flags in human-AI collaboration. Capture key insights on a shared board to help consolidate learning.
Assessment
The main group deliverable is the Task‑Delegation Matrix, which should be evaluated based on criteria such as the completeness of the task list, the appropriateness of delegation decisions (human‑only, AI‑assisted, full AI), the quality of justifications, evidence of AI testing, and awareness of potential risks.
Peer feedback provided during the gallery walk can be used formatively to support comparison and refinement of decisions.
An optional individual reflection may be used to assess how students’ trust in AI or delegation strategies evolved and how their decisions relate to the applied decision framework; this reflection is best graded on a pass/fail (or completion) basis, with qualitative comments from the teacher recommended to support learning. For advanced students, an optional extension task may involve creating a one‑page AI Integration Plan for a personal research idea using a more specialised AI tool (e.g., AutoML, domain‑specific LLM); this extension is not required to achieve the core learning outcomes.
Possible challenges
- Students may either over‑trust or under‑trust AI outputs when making delegation decisions.
- AI tools may produce hallucinations or factual errors during live testing.
- Time pressure may lead to inefficient trial‑and‑error prompting.
- Group dynamics may result in uneven participation.
How to adress them
- Use the pre‑class framework and in‑class facilitation to model calibrated trust and critical evaluation of AI outputs.
- Build in an explicit verification step by asking students to cross‑check at least one AI‑generated claim against a reliable source.
- Keep prompts simple and provide a shared prompt bank to reduce time lost to experimentation.
- Assign rotating group roles (e.g. scribe or devil’s advocate) to support balanced participation.
The activity can be adapted to different contexts and constraints. A non‑AI (paper‑only) version replaces live tool testing with theoretical justification based on literature on AI capabilities, placing greater emphasis on conceptual understanding rather than hands‑on use.
The activity can also be tailored to specific disciplines. For example, STEM courses may focus on code assistants for data cleaning and model building, Humanities courses on large language models for summarising literary criticism, and Health fields on AI tools for evidence synthesis. While the AI tools and risks may differ, the core delegation framework remains unchanged.
A fully online setting can be implemented using shared digital workspaces, breakout rooms, and screen‑sharing for the gallery walk.
For more advanced cohorts, an extended “design‑your‑own” version can be used in which, after the core lab, students apply the task‑delegation framework to a personal project (e.g. thesis idea), producing a more detailed integration plan over a longer timeframe.
Doiron, J.A./ Generative A.I. in Education (2024). Co-education: A Human-A.I. Collaboration Framework for Teaching and Learning [Video]. YouTube. https://www.youtube.com/watch?v=S_Mfa7qDeIY (accessed May 18, 2026).
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.
