wM-4_3: Integrating Chatbots in Course Design

A Practical Guide for Faculty at DCE

  

Are you thinking about adding a chatbot to your DCE course for students to use?

This guide reviews essential considerations, each with examples and advice, including:    

  • Should I use a Chatbot? Pedagogical Considerations  
  • Chatbot Platform Selection Considerations    
  • Configuring your Chatbot   
  • Privacy and Ethics    
  • Introducing the Chatbot to Students    
  • Frequently Asked Questions    
  • Links to Additional Resources

Please note: Tools, policies, and best practices are still evolving rapidly. If you find an inconsistency or want help walking through things, please feel free to schedule a Course Chatbot Considerations Consultation with the DCE Teaching and Learning staff.



1. Should I use a Chatbot? Pedagogical Considerations

Guiding Principle: Always use Goals-Centered Design Principles to decide whether a bot fulfills your course goals and learning objectives, just as you would for other learning assets.

What specific learning objective(s) will this chatbot help achieve?

Example learning objective: “To enhance comprehension and application of the Science of the Diversity Method."

What role will this chatbot play in the learning process?    

Tutorbot conveying course information

Assignment/ capstone assistant

Simulation/role play facilitator    

Practice problem generator

Simulation/role play facilitator    


How might the chatbot enhance student learning experience and outcomes compared to other types of learning interventions? What is the scope and limitation of this chatbot?

Personalized formative feedback, 24/7 availability, interactive practice, enhanced learning engagement and motivation Chatbot will limit its responses to course topics and redirect all other questions to faculty/ the teaching team


How do you want students to engage or interact with the bot (e.g. frequency, types of interaction)? Are there any behavioral guardrails you want to put in place?    

Chatbot is intended to provide collaborative problem-solving with learners    Students should interact with the chatbot 15 min per week.    Chatbot should adjust responses based on students' proficiency levels or prior interactions (scaffolding)    The chatbot should create friction to assist in students’ cognitive process (e.g. critical thinking and learning reflection)


How will the chatbot integrate with the existing course structure and content (e.g. learning activities and assessments)? Is the chatbot connected to homework assignments, class discussions, exams? What is your policy of student use of the chatbot during different learning periods (e.g. mid-term and end-of-term exam period)


What are the potential limitations or drawbacks of using a chatbot in this context? What might happen to the students’ learning experience, participation, or performance if something goes wrong? How can you mitigate these concerns?

e.g., over-reliance, inaccurate information, lack of human interaction, engagement and motivation?


How will you measure the effectiveness of the chatbot for meeting your course goals?

e.g., chat and engagement analytics, learning outcomes, teaching team’s observations on knowledge transfer and skills building, survey at the end of the course, question added to course evaluations


2. Chatbot Platform Selection Considerations

The currently approved tools (January 2026) are HUBot and PingPong. In both platforms, the instructor provides Instructions which tells the bot how to interact with students. The instructor can also upload documents with course content information to the platform’s Knowledge base.

Neither platform uses student input to train the LLMs. Any platform not approved by HUIT will need pre-approval from DCE’s IT teams to ensure it satisfies FERPA and security requirements.

Considerations
PingPong
HUBot
Do you need more than one chatbot in your course?
✅ Can create multiple bots per course
❌ Can only create one bot per course
Do you want students to be able to share threads (conversations with the bot) with you and other students?
✅ Students can share threads
❌ Students cannot share threads
How good is the Canvas integration?
↔️ Roster created from class registrants, but not synced to Canvas. Causes access issues if official vs. HKey emails don't match
✅ Installed directly in Canvas; students have automatic access
Does my bot need to be faithful by default to the knowledge base documents I’ve uploaded? (Note: All bots hallucinate and none can be exclusively restricted to its knowledge base.)
❌ Must use prompt engineering to tell bot to privilege knowledge base
↔️ Bot defaults to emphasize materials in its knowledge base
Will you allow the LLM to access to live Internet sources, beyond the trained data?
✅ Can retrieve new info from the Internet
❌ By default, no access to the live Internet
Is the user interface easy to use?
↔️ Some confusion. Instructors report the “temperature settings” are not intuitive.
✅ No confusion in UI reported
Can students talk to the Chatbot out loud?
✅ Offers Voice Mode
❌ No Voice mode
Who manages each platform? How do I get access? (Note: DCE does not manage access to either platform)
✅ Placeholder
Request access from the PingPong Team at the Harvard Kennedy School of Government.

Variations

1. Should I use a Chatbot? Pedagogical Considerations

Guiding Principle: Always use Goals-Centered Design Principles to decide whether a bot fulfills your course goals and learning objectives, just as you would for other learning assets.

Guiding Principle: Always use Goals-Centered Design Principles to decide whether a bot fulfills your course goals and learning objectives, just as you would for other learning assets.


Goals and objectives: What are the goals of the course and who are the potential users of this chatbot


Example of course goals and users: Help business professionals at Harvard Extension School apply sustainable financing concepts in their workplace


Example of high-level learning goals of the chatbot: knowledge recall, critical thinking, problem-solving, communication skills


What are the broad, high-level learning goals for this chatbot?