I ORGANIZED MY FIRST EVENT AS GEN AI CLUB PRESIDENT — PROMPT WARS

The Idea

Most students around me were using AI tools every day — ChatGPT, Gemini, whatever was trending — but very few were thinking about how they were using them.

The output quality depends entirely on the prompt. That's not a niche insight anymore — it's fast becoming a baseline skill across every field. Engineering, design, marketing, law. Doesn't matter. If you use AI, you need to know how to ask it things properly.

That felt like something worth competing on. So I built a competition around it.

What is Prompt Wars?

Prompt Wars is an AI prompting competition open to all students — no coding required, no prior AI background needed.

Three rounds, each testing a different dimension:

• Round 1 — Creative Image Generation — Given a theme, generate the best possible visual output using any AI tool of your choice
• Round 2 — Web Development with AI — Build something functional using AI assistance, judged on output quality and prompt strategy
• Final Round — Blind Challenge — A surprise task revealed live on the spot, in front of everyone

Any AI tool was allowed. ChatGPT, Gemini, Claude, Copilot — participant's choice. The only thing being judged was the quality of the prompt and what it produced.

Organized and conducted by: Gen AI Club, NIAT × Crescent

The Build-Up

I planned and ran this solo.

Everything from scratch — proposal writing, faculty approval, slot coordination, poster design, presentation, registration setup, judging rubric, event website with a live leaderboard, WhatsApp group for announcements. Each piece figured out one at a time, on top of regular classes and hostel life.

The slot situation was its own problem. Every club gets 1.5-hour windows. I needed 3 hours. Solution: two consecutive slots on April 16 — 8:50 AM to 11:50 AM. That worked.

The poster went through more iterations than I expected before it looked right. The proposal needed a full review cycle before approval. By the time April 16 came, every piece was in place.

On paper, at least.

What Happened on the Day

52 registered. 38 showed up.

I found out later why — people saw the registration count climb and quietly talked themselves out of attending. They had signed up thinking it would be a small, informal session. Once it looked like a proper event, some dropped. That's a framing problem. Next time, expectations get set from the first post.

The rounds ran well. Round 1 had people genuinely trying to outdo each other. Round 2 is where things got intense — participants were building real things under time pressure using AI. The Final Round had the kind of competitive atmosphere you can't manufacture — live, projected on screen, everyone watching.

Then evaluation hit.

What I had designed as a clean, timed scoring process became the most pressure-heavy part of the day. Judging manually, tracking scores, keeping time, answering questions — all of it at once, all on the same person. The interim leaderboard I had planned to display between rounds never went up. That was the clearest gap between what I designed and what actually happened.

Feedback

Most participants rated the event 4 or 5 out of 5.

A few responses that stayed with me:

• One participant said they built their first portfolio during Round 2 — not perfect, but a start. That line alone made the whole thing worth running.
• Someone noted the coordination was smooth and each round ran cleanly.
• A few asked for more time per round.
• One person specifically flagged evaluation transparency as something to fix next time.

Honest feedback. That's what I wanted.

What This Built In Me

• Evaluation needs a system before the event, not during it. Build the rubric, test it, assign scoring ownership early. Don't improvise this part.
• Display the leaderboard. Seeing scores in real time changes the energy in the room entirely. That feature existed — it just didn't go up. Won't happen again.
• Framing shapes attendance. Registration numbers affect who shows up. How you position the event from the first announcement determines who believes it's for them.
• Running something solo has a real cost. Every problem has one person's name on it. That's fine — but knowing exactly where the breaking points are is how you build something better next time.

Closing

Prompt Wars was the first event I ran. It wasn't perfect — the evaluation overran, the leaderboard stayed blank, and 14 people didn't show up.

But 38 did. They competed, gave honest feedback, and a few learned something they didn't know when they walked in.

That's a good first event.

The next one will be better.