The question has changed

A couple of years ago, the conversation in most schools was about adoption. Get teachers using AI. Normalise it. Stay curious. Lower the fear.

That still matters. But Dr. Shannon Doak, Director of Technology at Nanjing International School, makes a point in this week’s episode that I think marks a real shift in how schools need to be thinking: the responsibility has gotten harder.

“It’s not just use AI. It’s more about protecting thinking these days.”

Shannon has spent over thirty years in international schools across Guangzhou, Hawaii, and Nanjing. He’s watched this move faster than almost anything else he’s seen in edtech, and his concern now is not adoption. It’s judgment.

The TALK Framework

Shannon developed a framework called TALK to help teachers approach AI differently. It came out of a realisation he kept arriving at across workshops and presentations in Asia: teachers were treating AI like every other piece of technology. Something to learn. Something with a correct way of doing it.

His response was simple: just talk to it.

TALK stands for: Talk it out → Ask, explore, wonder → Listen → push back → Keep going → Create.

The framework isn’t really about AI. It’s about maintaining the teacher’s thinking throughout the process. AI performs better when you use it this way, Shannon says — and more importantly, so do you.

For students, the framework is almost identical. With one change: the final “Create” step belongs to the student, not the AI. They use the dialogue phase to expand their thinking. Then they put the AI aside and produce the work themselves.

A cleaner approach to assessment

One of the most useful things Shannon said in the whole conversation was about assessment — a topic that’s generating a lot of heat in schools right now.

His rule: if you’re assessing a skill, students can’t use AI for that specific skill. If you’re not assessing it, they can.

“If I’m trying to assess your ability to make an outline, you can’t use AI for that. But if I’m assessing your five-paragraph essay, making the outline — sure, go ahead.”

It’s not a total ban and it’s not a free-for-all. It’s a question: what are you actually trying to find out about this student?

On AI detectors

Shannon tested six of them. Within three prompts, he got 100% AI-generated content to pass as 100% human.

His advice: don’t use them to catch students. Use them to start a conversation.

“Even if it says 80% chance, you’ve got to let that go because it’s probably not true.” Especially, he notes, for ELL students, where the false positive rate is even higher.

The 80/20 rule

Something Shannon’s school has operated with from early on: AI does the heavy lifting — about 80%. But a teacher must put professional eyes on the work and own the final 20%. Not as a formality, but as a genuine quality check.

It’s a practical frame that respects both the tool and the professional.

One thing to do this week

Shannon’s takeaway for school leaders: help your teachers shift from using AI as a task doer to using it as a thought partner.

In practice, that means replacing prompts like “make me a lesson plan” with prompts like “challenge my lesson design” or “where is the cognitive demand too low?” or “what might I be missing here?”

Start with teachers’ existing thinking. Let AI expand it. End with teachers’ professional judgment. That’s the pattern.

Links

* Dr Shannon Doak’s website: https://www.drshannondoak.com

* Dr Shannon Doak’s Linkedin: LinkedIn

* Alignment Education: https://alignmentedu.com/

* Nanjing International School: https://www.nischina.org/

* IPLNA – International Professional Learning Network Association

* EdTechGZ 2025 Keynote: Look Who’s Keynoting

* Key articles:

* “Innovative EdTech: Wolf in Sheep’s Clothing” (LinkedIn Pulse, 2018)

* “Our Responsibility in an AI-Driven Future” (EdTechGZ Keynote, Jan 2025)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit braintoolsatschools.substack.com