EdReports finds a gap between AI feature claims and curriculum evidence

Curriculum directors, teachers and purchasing teams need to know what changed when a familiar product adds AI. Ask which feature is under consideration, whether evidence covers that version, and who can stop or override its output. Offer a bounded review of one curriculum renewal, mapping supplier claims to evidence and unresolved questions. The value hypothesis is a better-informed adoption decision with clearer conditions for a pilot. Do not promise learning gains, staff savings or certification from this scan. Its small market sample supplies useful discovery questions; it does not identify a customer's buying intent or establish that any individual product is ineffective.

For one proposed lesson-generation feature, document the full path from source curriculum through model output to teacher approval and learner access. Prerequisites include licensed content, test accounts, model-change information and curriculum reviewers. Use synthetic records to test cross-user isolation, unexpected external retrieval, answer leakage and failure recovery. A proposed proof of value should score sampled outputs against local standards, lesson sequence and accessible rendering, then measure teacher corrections. Require explicit permission boundaries for any multi-step agent. Do not infer deployment capacity or safe autonomous use from feature labels. Evaluate retained learning separately if the district proposes student-facing use.

Assign the curriculum owner responsibility for acceptance, with IT, privacy, procurement and accessibility reviewers. Start with a feature inventory and a small set of teacher-reviewed lesson scenarios; reserve professional-learning time for judging and correcting outputs. Dependencies include supplier disclosures, approved data use and an accessible fallback.

Proposed acceptance

every pilot feature has an owner and version record, all sampled lessons receive curriculum review, and unresolved critical defects block expansion. Track correction time and student experience alongside any later independent assessment. These are proposed local criteria, not observed results. Risks include silent supplier changes and review work exceeding available teacher capacity.