Competence domain of the Didactic Framework: Creative and Practical Application

By the end of this activity, students can… 

• communicate and interact with GenAI effectively (formulate clear and effective prompts, iterate prompts and refine outputs). (FLAIR Didactic Framework: LO16) 
• analyze communication breakdowns that occur when collaborators operate with different implicit rules or assumptions. 
• explain why current AI tools present specific collaboration challenges, such as limited memory, system-level constraints and restricted self-awareness. 
• apply theory of mind concepts to human-AI interaction by recognizing asymmetries in knowledge and assumptions between humans and AI systems. 

Students play the card game Barnga in small groups of 4-6, following table specific rule sheets without verbal communication. As players rotate between tables, they encounter conflicting rule systems, leading to confusion and frustration. The experience is deliberately left unexplained until the debrief. Detailed instructions for running Barnga are provided in the Further Resources section. 

Bridge to theory of mind and human-AI interaction (10 min)  Introduce the concept of theory of mind, emphasizing that understanding others requires recognizing differences in knowledge, beliefs, and perspectives. Build on the concept by explaining the four-way gap relevant to human-AI interaction: AI systems do not know the user, users do not fully know the AI, users may not always know their own assumptions and needs, and AI systems do not fully know their own capabilities and limitations. 

Link between Barnga and AI collaboration (10 min) 
Explain why the Barnga experience directly maps onto collaboration with AI. Point out that AI tools are often designed to execute rather than ask clarifying questions and illustrate this using an excerpt from a system prompt (e.g. Claude – see Further Resources section). Emphasise that AI tools typically do not retain memory across conversations, meaning there is no accumulated shared context or mental model of the user (note: memory features emerging in 2026 have limited efficacy so far). Students shall understand that similar to switching tables in Barnga, each new interaction with an AI system represents a fresh start with potential for misaligned assumptions. 

Getting to know AI (10 min)  Discuss strategies for reducing the knowledge gap in human-AI interactions, such as: 

• Using AI tools repeatedly to build intuition 
• Reading evaluations and benchmarks to understand capabilities 
• Examining system prompts to better understand AI behavior (explain what system prompts are, share an example and suggest students check periodically for updates) 

Prompting strategies (10 min)  Introduce practical prompting strategies that help reduce misalignment and improve collaboration with AI. Here are some prompting techniques from Claude’s documentation: 

• Be specific and provide context 
• Break complex tasks into steps 
• Ask AI to explain its reasoning 
• Iterate and refine based on outputs 
• Provide examples of desired outputs 

Wrap-up and Q&A (5 min)  Summarize key takeaways. Distribute a handout with the system prompt excerpt and useful links to prompting resources. Invite questions and encourage students to experiment with these strategies and reflect on them in their own AI use – in class, or later. 

Possible challenges

• The connection between Barnga and AI collaboration may not be immediately obvious to all students. 
• If Barnga rule sheets are not carefully managed, students may compare rules too early, which reduces the impact of the activity. 
• Some students may experience genuine frustration or irritation during the Barnga card game. 

How to adress them

• Make the link between the Barnga experience and human-AI collaboration explicit during the debrief by using concrete examples from AI tools and interactions. 
• Collect all rule sheets before gameplay begins and avoid explanations while the game is in progress to preserve the effect of differing implicit rules. 
• Acknowledge emotional reactions during the game and frame frustration as valuable data for understanding collaboration under uncertainty. 

• Barnga game materials and instructions: https://sites.lsa.umich.edu/equitable-teaching/barnga/  

• Barnga game in action (video): https://www.youtube.com/watch?v=wp_S1NDafRs  

• Prompt engineering tips by Anthropic: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/overview  

• System prompts by Anthropic: https://platform.claude.com/docs/en/release-notes/system-prompts   

Excerpt from Claude Opus’s 4.5 system prompt (November 24, 2025): “…In general conversation, Claude doesn’t always ask questions but, when it does it tries to avoid overwhelming the person with more than one question per response. Claude does its best to address the person’s query, even if ambiguous, before asking for clarification or additional information…”