Teaching Innovation
Conversational Pedagogy
Good content doesn't create engagement — active, social learning does. During the lockdown I turned an ordinary chat app into a classroom where the students became the engine of learning, and I was the intelligence driving it.
When the lockdown moved everyone online and much of academia was struggling with the new medium, I didn't reach for a learning-management system. I built my classroom inside Slack — a natural chat interface — and ran it with a handful of simple activities and an assessment platform (Examineer) my team had built. No portal, no course-ware; just a conversation.
Every activity was designed to do two things at once: pull each learner into the content, and pull learners into conversation with each other. I mostly played catalyst — adding a pinch of perspective and keeping the room alive. Here is what I did, and what I learned.
No LMS, by design
A learning-management system runs a course; it doesn't make anyone want to show up. LMSs are built for content and user management — not engagement. Slack is the opposite, an engagement tool at heart, so I used it to run the course and got the engagement for free. A handful of purpose-built channels did everything an LMS manages, in a space people actually enjoyed.
Course announcements. Instructors post; students react with emoji.
Instructors post — and each post becomes the home for one short video or a self-contained lecture-note PDF. Students watch or read right there and drop doubts in the thread beneath, so every lecture gets its own discussion. (Videos were kept deliberately short; the PDF tied them together and stood on its own.)
One set a week, with the round-robin solution assignment. Everyone posts, replies in threads, and reacts.
Home for the engagement activities — Concept Hunt, Build Your Own Exam, and the rest.
A private channel for each project team to plan, share documents, and build together — which also let me watch collaboration happen in real time and gauge everyone's involvement.
Everything else — a place to hang out, introduce yourselves, float ideas, and pull each other's legs. Open to all.
To warm the room up — and make sure everyone actually showed up and got acquainted — I'd open each course in #lounge by asking students to introduce themselves with their hometown and its most famous food or snack. Food connects people: it got everyone talking, and it made the teaching staff feel human rather than distant. For the shy ones hanging back, I'd call them out by name and ask their friends to nudge them — half the fun was watching classmates poke each other to speak up. The message was gentle but clear: everyone matters here, and hiding in the crowd isn't an option.
Formal assessments and submissions ran on Examineer, the assessment platform my team built. Everything else lived in the conversation.
The activities
A few that any teacher can borrow. (More to come as I write them up.)
Concept Hunt
Before I taught a concept, the class discovered and explained it themselves.
A few days before a topic, I'd post a 2–3 line description and one short, visually interesting YouTube clip — under five minutes, the shorter the better — introducing it. Then I'd set the class loose: find and share your own short videos on the same concept, and say, in your own words, why you found each one interesting.
The twist made it fly: only the first ten new videos earned points — no repeats — so being first, and finding something nobody else had, carried a small thrill. Within minutes the timeline would fill up; a class of ~100 routinely gave me fifteen videos when I asked for ten. I'd add an emoji legend and ask everyone else to rate each video, comment, and ask questions — and, remarkably, I almost never had to answer anything myself. The students answered each other.
Within minutes the class turned from an audience into contributors — hunting, explaining, rating, and answering each other's questions. Being first to find something new made discovery a game, so the students, not me, kept the room alive.
Crowd-Solved Problem Sets
Every problem got two student-written solutions, crowd-reviewed.
Students did the weekly set as usual — but I also assigned two students to each problem, round-robin across the semester. Those two each posted a clean, typed-up (LaTeX) solution. The solutions went up for the class to comment on, and the authors revised in response.
The payoff was rich: every problem had two solutions to compare, peer feedback sharpened them, and real discussion erupted on the hard ones — a gold mine for a teacher. Solutions were crowd-reviewed and approved, with my expert insight added only where it was needed. And I never had to write up a single solution myself.
Everyone was an author, not a spectator. Public, side-by-side solutions pulled the class into critiquing, defending, and refining — and the hardest problems drew the liveliest debate.
Build Your Own Exam
Students invented original problems for the exam — and half the final came from them.
I asked students to create the kinds of problems they'd want to see on the final — each one novel and original, not lifted from a textbook or the internet — and promised that about half the exam would be built from their submissions. To submit a problem you first had to solve it — but you didn't submit your solution. Then volunteers solved the submitted problems and posted typed-up solutions, and could challenge a problem's accuracy or wording, or offer an alternate perspective. Two-plus solutions per problem.
By the end, every student had a bank of peer-authored, crowd-reviewed problems and solutions to study from — a guaranteed baseline of preparation for everyone. I received several genuinely good problems, refined them, and put them on the actual final. And the deepest benefit was subtler: to invent an original problem you have to understand a concept from the inside — so writing questions sharpened their physics and their creativity at once, and gave them real ownership of the exam.
With their own problems on the line — literally on the exam — students engaged like stakeholders. Inventing, solving, and challenging each other's questions turned exam prep into a creative, collective effort.
Team Projects, in Team Channels
Every team got its own channel — and the work happened where I could see it.
I split the class into teams and gave each one a dedicated channel to plan, share documents, and build their project together. Because the collaboration lived in the open — at least to me — I could watch it unfold in real time: who was contributing, how the team was dividing the work, where they were stuck. It doubled as their shared workspace and let me gauge each student's involvement without ever having to ask.
A team channel turned a project into a shared, visible workspace. Working in the open kept everyone accountable and involved, and gave each team a natural home to think together.
What made it work
- Two kinds of engagement, always. Each activity pulled a learner into the content and into conversation with peers — never one without the other.
- Students as the engine. Discovery, a little competition, ownership, peer teaching, and crowd review did the heavy lifting — not me lecturing.
- Formative, not final. Assessment lived inside the learning — priming, solving, reviewing — so it shaped understanding instead of just measuring it.
- The teacher as catalyst. My job was to spark and steer, add perspective, and keep the room alive — not to be the sole source of answers.
- A conversation, not a platform. It all lived in a chat window. No LMS — and engagement was higher, not lower, for it.
Where it goes
What I ran was instructor-driven blended learning: I was the intelligence driving engagement at both the individual and the cohort level — by hand. That is exactly the part a capable AI can now take on.
Turn the human driver into an AI one, and this becomes technology-driven blended learning — an AI coach sparking the same discovery, discussion, and ownership for every learner and every cohort, around the clock. It's the direction I've been most excited about for years.
I first sketched this direction in an earlier essay — read it on Medium →