In Short:
Analog first, AI second. A quick chalkboard mind-map grounds definitions and strategy before any screen lights up.
Tech as an accelerator, not a crutch. AI speeds the process while keeping performance and real-time thinking squarely in students’ hands.
Faced with the motion “Is learning English still relevant in Hong Kong today?” a secondary debate team skipped the usual scatter-shot Googling and trialled a lean AI workflow. One brisk session—analog brainstorm, targeted deep research, auto-outlining, and text-to-speech rehearsal—produced draft speeches for every speaker. By shifting grunt work to machines, students spent their energy on strategy, rebuttals, and stagecraft.
How It Worked
Activity: Team huddled around a chalkboard, unpacking definitions, burdens, and stakeholders. Phones stayed in pockets; a quick photo became the plan.
No AI
Activity: Pairs spent 15 minutes with Perplexity Research Mode and Gemini Deep Research to surface statistics, expert quotes, and counter-examples. Each pair shared 1-2 key findings with the team.
Prompt: “You are a research assistant for a school debate. Motion: Is learning English still relevant in Hong Kong today? 1. List the three strongest claims for and against (≤40 words each). 2. Provide three credible sources per claim (title, year, link). 3.Flag key rebuttals. Return as a table.”
Activity: A Voice Memo captured a 20-minute strategy discussion. Their discussion was transcribed and then organized into a clear outline. Breaking down who will talk about which points using which evidence and what questions they can raise.
Prompt: “Turn this transcript into a concise outline: headings = argument, bullets = supporting evidence, note any action items.”
Activity: Each speaker merged the outline and evidence into a 4-minute script inside ChatGPT. They had to read it out to themselves and edit it (with AI or otherwise) so that it is easy and smooth for them to read out loud.
Prompt: “Draft a persuasive debate speech using our outline. Clear signposting, conversational tone, memorable closing line. Keep it under four minutes.”
“I added this speech to use simpler language. I prefer shorter sentences and repetition of these keywords…”
Activity: I demoed OpenAI.fm to show how text-to-speech playback can highlight pacing, stress, and pronunciation; students later used it for individual practice.
(Tool only)
Impact
“We went from blank page to a first-draft speech in one meeting. That never happens.”
“Because an AI note-taker had our backs, we could focus on testing arguments instead of scribbling minutes.”
As a coach, I loved that the usual “Who’s got evidence for…?” scramble was much diminished; AI-generated evidence tables were instantly available, and made students feel more secure and equipped to speak out. Freed from procedural clutter such as formatting and diarising, the group spent its energy poking holes in each other’s logic and rehearsing delivery. Rapid progress raised engagement—students left buzzing, eager to refine rather than start over.