Faculty Projects
Community-Centered AI: A Responsible Course Assistant Pilot on Ed Discussion

Lakshmy Mohandas and Yiji Zhang headshots

Faculty

Lakshmy Mohandas, UEO/Segal
Yiji Zhang, CS

Amount Requested

$5,070

Summary

Generative AI tools are already part of how McCormick undergraduates learn on a daily basis, but students' unguided usage creates three risks for their undergraduate experience: isolation from the course community (classmates, peer mentors, TAs, and the instructors), declining attendance at office hours, and unverified answer quality. Banning AI is neither realistic nor desirable. The open question is how to design AI assistance so that it strengthens both the community and students' learning.

In partnership with Ed Discussion, a modern Q&A platform already used in many engineering courses across the country, this project aims at piloting a controlled, instructor-configured AI Assistant embedded directly in the course community.

The bot will do more than answer quickly. In keeping with McCormick's strategic pillar of Transforming Engineering Education and the school's leadership in responsible, human-centered AI, we will deliberately configure and study how an AI Assistant can (1) strengthen, rather than replace, peer-to-peer discussion; (2) actively route students toward office hours and human help; (3) signal that instructors want to talk with students;(4) help identify Frequently Asked Questions (FAQ's) and (5) maintain rigorously measured answer accuracy. We will pilot in two courses in Fall 2026, analyze results in Winter 2027, and share findings and an adoption playbook with the McCormick and Northwestern community in Spring 2027.

Planned Activities/Investments

September 2026 (preparation):

  • Finalize collaboration terms with Ed Discussion;
  • design bot configurations (e.g., responses that cite course materials, suggest relevant classmate threads, and recommend specific office hours);
  • develop accuracy-evaluation rubric
  • develop IRB protocol for student data.

Fall 2026 — Pilot in three courses:

  • Deploy the configured AI assistant in CS380 Software Quality Engineering (Zhang), Design Thinking and Technology (DTC) (Mohandas), and the McCormick First-year Academic Advising course (Mohandas).
  • Test community-building configurations: the bot links students to related peer discussions, defers to humans on judgment questions, and explicitly invites students to office hours with instructor-written messaging ("I'd like to hear more about this. Come by Tuesday").
  • Log response accuracy: trained undergraduate student research assistants audit a weekly sample of bot answers against course materials, with instructor spot-checks.
  • Collect data: office-hours attendance, Ed engagement metrics (peer replies, thread participation), and brief student surveys on belonging and connectedness, compared against prior offerings of the same courses/experience.

Winter 2027 — Analysis:

  • Analyze accuracy audits, engagement and attendance data, and survey results.
  • Identify which bot configurations increased (or harmed) community indicators.

Spring 2027 — Adoption:

  • Publish a practical configuration guide ("playbook") so any instructor can adopt the approach;
  • Work with interested departments and Northwestern IT/Ed Discussion on broader rollout
  • Host a McCormick workshop for faculty in collaboration with either NCEER, Searle Center, or NU IT

Impact

AI-Assisted Learning in Technical Courses

82 undergraduates in two Fall 2026 courses- Prof Zhang's Software Quality course and Prof Mohandas's DTC course - will receive fast, accuracy-audited AI assistance embedded directly within their course community on Ed Discussion, rather than through an isolated external chat window. This integration means students get help in context, within the same space where they are already asking questions, collaborating with peers, and engaging with course content.

First-Year Advising and Community Building

Professor Mohandas advises 90 incoming McCormick freshmen, a critical population whose early experiences significantly shape their long-term success, sense of belonging, and retention in engineering. A dedicated Canvas advising course will be created for these students, incorporating Ed Discussion as a community and information hub. This space will serve multiple purposes simultaneously:

  • Community building: Incoming freshmen will be able to connect with one another, share experiences, and build peer relationships before and during their first year. This also helps with student retention and well-being.
  • Peer advising network: Students will be able to see who their peer advisers, major-specific ambassadors are and connect with them directly, expanding their support network beyond faculty.
  • On-demand advising information: The Ed Discussion AI bot will be loaded with frequently asked questions covering key logistical information students need, such as drop deadlines, important contacts, course recommendations, and academic planning guidance, available instantly at any time of day.
  • Focused advising hours: By offloading routine logistical questions to the AI bot, Professor Mohandas's advising hours can be redirected toward meaningful personal, academic, and professional development conversations with students, rather than repeatedly answering the same administrative questions, allowing advising time to be used where it matters most.

This module can also be used for incoming Inter-School Transfer (IST) students to connect with peers and mitigate some of the learning curves they face transferring to McCormick.

Because Ed Discussion is already used at McCormick, successful configurations can be adopted by other courses within the same academic year.

Strategic alignment-wise, we hope to advance the Transform Engineering Education pillar by modernizing how students get help, and model McCormick's whole-brain, human-centered approach to AI technology in service of community.

We also hope to produce practical, evidence-based guidance on a question every engineering school is currently facing, positioning McCormick as a leader in responsible educational AI.

Deliverables

  • A working, instructor-configurable AI assistant deployed on Ed Discussion in two McCormick courses (Fall 2026) and an AI-enhanced advising hub on Canvas for 90 incoming freshmen, both launching Fall 2026.
  • An accuracy-audit dataset and report on bot response quality.
  • An evaluation report: effects on office-hours attendance, peer engagement, and student sense of community.
  • A public adoption playbook: recommended configurations, instructor messaging templates, and setup guide for McCormick faculty.
  • A Spring 2027 faculty workshop and presentation of results to the McCormick community.

Sustainability

After the pilot, ongoing costs are limited to LLM API usage, which individual courses or departments can support. The playbook and configurations are reusable at no additional cost, and Ed Discussion partnership terms will be negotiated with continuation in mind.

Budget Overview

  • Undergraduate Student Hires (2 students, data collection & analysis): $3,200 — Two undergraduate students hired at $20/hr for 8 hrs/week over 10 weeks (one quarter). Provides two undergraduate students with hands-on experience in AI evaluation, data analysis, and educational research - highly valuable skills for engineering and technology careers.
  • LLM API usage: $120 — Covers OpenAI GPT-4o API token costs across 3 courses (172 students total) assuming 50 questions per student per quarter.Enables 172 students to receive instant, 24/7 AI-assisted responses to academic and advising questions, reducing wait times and supporting learning outside class hours.
  • Survey instruments and student participation incentives in surveys and focus groups: $750 — Covers refreshments and other incentives to encourage participation in surveys and focus groups. Surveys and focus groups use Qualtrics and printed materials, respectively. Encourages honest student feedback, ensuring the study captures the true undergraduate experience and leads to improvements benefiting future students.
  • Spring workshop and dissemination materials (printing, refreshments, playbook production): $1,000 —Covers refreshments for end-of-year faculty/staff workshop ($300), production and printing of a practical adoption playbook ($400), and poster/presentation materials ($300). Ensures findings are shared broadly across McCormick, so improvements benefit not just the 172 students directly served but undergraduates across the institution.

Total Budget Amount: $5,070

Matching Funds

Undergraduate Engineering Office, Segal and CS dept