Executive Summary (Download as PDF)
In December 2025, the TeachingAbout.AI project convened school librarians, classroom teachers, instructional leaders, university researchers, and instructional technologists in Rochester, New York. Over four days, the panel set aside discussions about AI implementation and teacher training to consider a harder question: what should K–12 educators be asking before we begin to tackle AI work? The answer took the form of eight deliberately uncomfortable statements published as the Rochester Provocations. The guiding frame was taken from Neil Postman’s work on technological change. A new technology is never simply added to an environment, Postman noted, instead it changes it on an ecological level. The arrival of generative AI has not left us with the old school system plus a chatbot; instead, it is recoloring every interaction between teacher and student, student and text, and school
and community.
This reframes AI in education as a wicked problem with no right or wrong answers, only better and worse ones that change as we learn. The convening’s central conclusion follows: the deepest issues AI raises around cheating, the validity of assessment, the role of the teacher, the purpose of homework, and the conditions for teacher learning are not actually new problems caused by AI. They are old, unresolved problems AI has made impossible to ignore. AI is the catalyst, not the cause. Treating AI as a self-contained problem invites a self-contained “solution”—detectors, bans, locked-down browsers—that substitutes a purchase decision for a pedagogical one and changes almost nothing.
The provocations point instead toward redesign: leading with assessment validity rather than policing cheating, abandoning the alluring impossibility of “AI-proof,” naming what only human teachers can do, granting teachers explicit permission to iterate, and meeting AI’s documented harms with harm reduction rather than avoidance. None of this is new. These are changes thoughtful educators have argued for since long before generative AI. What AI adds is urgency. This work will succeed or fail in the hands of three roles in particular, classroom teachers, building principals, and school librarians. School librarians, already the closest thing most districts have to AI-fluent instructional leaders, belong at the center of the conversation. The priorities below translate the provocations into concrete starting points for each role.
For Classroom Teachers
- Move from punishment to a restorative protocol for suspected AI misuse: conversation, redo opportunity, explicit teaching about the purpose of the assignment.
- Pilot Black Box Assessment in two or three units per grade band where the work has traditionally been a take-home product. Capture and credit the rough draft, the revision, and the rationale—not only the final paper.
- Use AI feedback as preparation for human feedback, not as a substitute. A chatbot can scaffold a first draft; only a teacher can tell a student what the work means and provide recognition.
For Principals and Instructional Leaders
- Reframe district planning documents from “AI policy” to “teaching and learning in an AI-mediated context.” The shift is not cosmetic; it changes what counts as a relevant solution and who needs to be in the room.
- Issue explicit, written permission from district leadership: this is the year we redesign. Specify what divergence is allowed, what evidence is requested, and what supports are available.
- Replace “AI ban” and “AI mandate” framings with explicit harm-reduction policies. Name the harms (engagement-driven design, parasocial bonding, deepfake abuse) and the mitigations.
For School Librarians
- Empower school librarians as in-house consultants on task redesign. Their co-teaching role across subject areas makes them well placed to coach colleagues through structural redesign one unit at a time.
- Use the LibraryReady.AI PreK–12 scope and sequence as a backbone. It is grade-banded, librarian-friendly, and built for exactly this work in all levels of classrooms.
- Treat school librarians as the in-district R&D unit. Their broad view across classrooms, their information-evaluation skills, and their relationship to the LibraryReady.AI scope and sequence make them ideal redesign partners.
The TeachingAbout.AI K-12 Field Report was produced as part of a grant funded through Google. It is released under a Creative Commons BY 4.0 License. Project lead and primary report author: Dr. Christopher Harris, Director of Libraries and Digital Learning Services at Genesee Valley BOCES.