In September 2024, I ran a 2-day Faculty Development Program on AI Tools for Education at a mid-sized engineering college in Karnataka. 40 faculty members. Mixed departments — CS, ECE, MBA, and a few from Physics and Maths. Most had used ChatGPT for something, many only barely.
Here's how we designed it, what actually happened, and what I'd change next time.
The Brief
The HOD who reached out had a specific ask: "Our faculty need to understand AI tools — but not just ChatGPT. We want them to actually use these tools for their teaching work, not just learn about AI as a topic."
That framing shaped everything. This was not an "introduction to AI" FDP. It was a workflow change workshop. The goal was: every faculty member leaves with a tool they'll use from the next working day.
We had 16 hours spread across two days. 40 participants. A mix of departments, age ranges (25 to 55+), and technical comfort levels.
Day 1 Design: Start With the Pain
I've learned that if you start with the tools, you get resistance. "I don't have time to learn another app." "Our students will misuse this."
Instead, I start Day 1 with a 20-minute discussion: what are the most time-consuming parts of your current job?
The answers are always the same:
- Writing question papers
- Preparing slides for a new topic
- Marking assignments and writing feedback
- Reading research papers (for PhD faculty)
- Drafting circulars and communication
Once the pain is on the board, I pull up the tools. Suddenly it's not "a new app to learn" — it's "a thing that solves the problem we just listed."
What Actually Happened in the Classroom
The moment that changed the room: a Physics professor had a unit on Quantum Mechanics that she described as "notoriously hard to make interesting." I pulled up Claude, dropped in the learning objectives, and in 90 seconds generated six different analogy-based explanations at different abstraction levels, a 5-question formative quiz, and a 3-slide outline.
Her reaction — and the visible ripple through the room — was worth more than any introductory lecture I could have given.
By the lunch break on Day 1, every single participant had generated at least one artifact they said was genuinely useful.
The NotebookLM Session
This was the surprise hit of the FDP. I expected the LLM chatbot sessions to get the most engagement. NotebookLM outperformed everything.
Two reasons:
- It's non-threatening. You upload documents you already have — your own notes, a textbook chapter, NPTEL PDFs. It's not "AI generating content" — it's AI helping you understand content you're responsible for.
- The Audio Overview feature. When I showed a faculty member that NotebookLM could generate a 10-minute podcast-style discussion of their uploaded research paper, there was a genuine pause of disbelief.
What Didn't Work
Gamma.app on slow college WiFi. We knew this might be an issue and brought a mobile hotspot. Still, the cloud-rendered presentations were slow to generate during the session. Next time: preload demos, don't rely on live generation for time-sensitive sessions.
The "academic integrity" detour. I had a 30-minute slot for discussing AI tool policies with students. The discussion got heated (in a good way — a lot of genuine debate about what's fair), but it ran 45 minutes and compressed the hands-on time for Module 4.
Lesson: Put the policy discussion at the end of a day, not the middle.
The tech confidence gap was wider than expected. About 8 faculty members in the ECE and MBA departments had never used a browser extension or copy-pasted a prompt.
Fix: Pair programming style — assign one tech-comfortable faculty member as a "buddy" to every less-comfortable participant for the tool setup sessions.
Day 2: Building Personal Workflows
Day 2 was more applied. Each participant mapped their own top 3 repetitive teaching tasks and spent time finding the specific tool/prompt combination that addressed each one.
The structured "workflow documentation" exercise — where participants wrote out their before/after workflow — turned out to be the thing that made the learning stick.
Outcomes We Could Measure
- Time saved (self-reported): Average participant reported 3–4 hours/week of potential time savings
- Tools adopted by end of day 2: Every participant had set up at least 2 tools; 70% had used at least one to generate a real artifact they planned to use
- 3-week follow-up (informal): 8 of 40 participants reached out directly to share artifacts they'd made
What I'd Do Differently
- Send a pre-FDP survey. Collect: what subjects do you teach, what are your 3 most time-consuming tasks, have you used AI tools before?
- Offline-first demos. Cloud tools are unreliable on college WiFi. Preload every demo.
- Leave time for "custom problem" sessions. By Day 2, participants had their own problems they wanted help with.
Running FDPs for faculty is some of the most impactful training work I do. The multiplier effect is real — 40 faculty members who are AI-fluent are in rooms with thousands of students every week.
If you're a TPO or HOD thinking about whether an FDP makes sense for your faculty, reach out here — happy to walk you through what a customised session would look like for your team.