Summer Kickoff: A 90-Minute Tour of AI at S2
From "I've heard of ChatGPT" to "here's where AI fits in my work, here's where it doesn't, and here's one specific change I'm going to make next week."
Facilitator toolkit
Everything you need to teach this sessionSlide deck → present
29 slides, keyboard nav, fullscreen, speaker view, timer. Project this on the main screen; alternate with Claude for demos.
Facilitator notes
Block-by-block run-of-show with watch-fors, demo checklists, pre-flight, and the recovery plan if you fall behind.
Bayside files
Lice email, Field Day budget, Brennan deck, Ms. Patel CLAUDE.md. Drag into Claude live during the demos.
Pre-session survey
Three questions teachers fill out 24 hours before. Submissions land in a Google Sheet via Apps Script.
Live results dashboard
Aggregated survey responses, refreshes every 10s. Project this during the opening poll moment.
Pick-one pledge
Where teachers commit to one change. Five fields, recorded centrally in the Sheet plus a copy-to-clipboard summary they email to their partner.
Lobby screen
Project this while people arrive: big survey URL, QR code, a live counter, and a wall of who's already done it.
Live poll control
Your in-house Mentimeter. Push a question, every phone switches to it. Multiple choice, word cloud, scale. Anonymous.
Live poll results
The big-screen results: bars, histograms, word clouds, updating live. Shows the join code when paused.
Printable handout
One-page summary of the Five Levels, disclosure, red lines, and the pledge prompt. Print or save as PDF.
Backend setup
Wire the forms to a Google Sheet via Apps Script. Step-by-step walkthrough with copy-paste code.
Session recording
After your first session, drop the recording URL into the materials table below. Embed slot ready.
Why this session
About three in ten American teachers now use AI weekly, saving roughly six hours a week on tasks like preparing for class, drafting worksheets, and adapting materials for different learners. At the same time, a Canadian airline was forced by a tribunal to honor a refund policy its chatbot had invented, and a California lawyer was fined $10,000 last year when 21 of the 23 case quotes in his appellate brief turned out to be ChatGPT fabrications.
AI is a powerful, fast-moving tool. It is often confidently wrong. And it is becoming part of the work we do. This session is the practical overview every S2 teacher gets before AI shows up in their classroom and their inbox.
What you'll leave with
A plain-language read on AI
How LLMs actually work, why they make things up, and where they fail.
Where S2 stands
What the staff and student policies say and how they land in your day-to-day.
One specific change
Named task, named AI move, named accountability partner. Two-week check-in.
Run of show
-
10:00 — 0:10 · 10 minutes
Welcome and opening poll
A live word-cloud question (on the in-house poll) for "one word: how do you feel about AI right now." Three further survey questions (current usage, comfort 1–10, one task you'd love AI to take off your plate) sent in advance and aggregated on screen. Short frame on why this session exists. Format preview: live, single screen, audience reacting throughout, laptops closed.
-
20:10 — 0:20 · 10 minutes
Under the hood
A plain-language tour of how LLMs actually work. Five fast beats: predict the next word, meaning is a place in coordinate space, the neural net as an assembly line of simple math, how it learned by guessing billions of times and getting nudged toward right, and where and why it fails.
-
30:20 — 0:35 · 15 minutes
The S2 AI policy in practice
The policy is set. This block walks through what it says, why we landed where we did, and what it means for daily classroom practice. Disclosure-first posture. Grade-band expectations. Protected assessments. Privacy and data-handling guidance. Discussion welcome on how the policy lands in specific subjects and grade levels.
-
40:35 — 1:05 · 30 minutes
Five Levels of AI in Action
The heart of the session. Demos move from the floor to the ceiling, all set at the fictional Bayside Charter School. Each level is more powerful than the last, and the lens that ties them together is who is driving. In the early levels, the human is in the loop. By the last level, the human is on the loop. Closes with a short "how AI fails in 2026" beat, demonstrated live.
-
51:05 — 1:15 · 10 minutes
Pick one and find your partner
Each teacher writes down one specific change they will make in their work over the next two weeks, naming the task and the AI move they will use. Pairs are pre-assigned by the facilitator (mixed subject, grade band, comfort level). Each person shares their change, partner asks one sharpening question, both schedule a 15-minute check-in on their calendars before standing up. Three pre-selected pairs report out.
-
61:15 — 1:30 · 15 minutes
Closing showcase
One honest example of AI woven into real work: the facilitator shows that this whole session (deck, demos, handout, site, poll) was built with AI. Five minutes to close: where the policy lives, when the next session is, what to bring to the school-year drip. When a teacher leader is recruited, this is the slot for their classroom move.
The Five Levels — quick tour
These are the five levels demonstrated in Block 4, all set at the fictional Bayside Charter School. Each card links to a deeper write-up in the Five Levels reference.
Chat
Cleaning up Principal Henderson's panicked lice email. Same prompt produces the Spanish version.
AI works with you
Cowork audits Bayside's Field Day budget. Cowork rewrites Mr. Brennan's Back-to-School Night deck.
AI knows you
Custom GPT, Skill, and Memory — three flavors of "AI that knows your context."
Build with AI
A small working tool made from scratch with a handful of lines of code. The point is the door, not the tool.
Agents
An AI that watches attendance, flags students at risk, and drafts personalized outreach for your review.
Materials
| Item | Format | When delivered |
|---|---|---|
| Pre-session survey (4 questions) | In-house survey → Google Sheet | 1 week before session |
| Live in-class poll (word cloud, choice, scale) | In-house poll, on teachers' phones | In session |
| Run-of-show + facilitator notes | Markdown / PDF | This page |
| Live single-screen environment | Claude Desktop + Cowork | In session |
| Bayside Charter demo assets | Sample files | Coming to Bayside Demo Vault |
| "Your first ten prompts" handout | End of session | |
| "Pick one" pledge card | Print + digital | In session |
| Session recording + slide PDF | Video + PDF | Posted within 24 hours |
Format and logistics
One screen. Live tool, not slides. The built-in live poll (poll-control + poll-present) is open for the opening poll and any live question, in place of Mentimeter. The facilitator drives all the demos. 25 to 30 teachers per session, delivered per building or all-school combined. An optional 15-minute stick-around after the session for anyone who wants to try one of the prompts on their own material before leaving.
Status and next steps
| Working group review | In progress |
| Policy finalization (prerequisite) | Targeting Aug 15 adoption |
| Teacher leader recruited | Open · not required for the Kickoff (facilitator demos solo for now); slot ready when recruited |
| Summer dates locked with principals | Open |
| Bayside demo assets built | Open |
| Pre-session survey launched | Open · 1 week before each session |