Syntheses

• AI chat transcripts can make student thinking visible

• AI grading and feedback systems need validation before schools trust them

• AI literacy and assessment integrity need separate workstreams

• In an AI world, schools need visible thinking, not just policing final products

• District AI implementation needs an operating model, not just a tool rollout

• AI literacy needs a mix of interaction contexts

• Persona-based AI needs purpose, boundaries, and reality checks

• Constructionist AI literacy means students learn AI by building and testing things

• From district redesign to classroom discipline

• From observable thinking to validated claims