District staff urges AI literacy, privacy safeguards as Bloomfield Hills pilots Magic School AI

Dave Shulkin, a district staff member, told the Bloomfield Hills Schools Board of Education on Oct. 27 that artificial intelligence should be treated as a tool that augments—rather than replaces—classroom teaching.

"AI is nothing more than this. It is a pattern recognizer," Shulkin said during a 40-minute presentation that framed AI as a set of machine-learning processes, described common industry examples, and called for education-focused literacy and oversight.

Shulkin urged board members to consider AI from two perspectives: the educator/board viewpoint and the human/parent viewpoint. He outlined a district roadmap that started with initial experimentation after the launch of early generative models in late 2022, pilot work in 2023, district guidelines issued in February, and readiness and engagement steps for 2025–27. He said the district's planning groups had developed a vision and three "buckets" for guidance covering students, teachers and district operations.

Staff emphasized both benefits and limits. Shulkin described classroom uses—rapidly customizing materials tied to IEP accommodations, producing short scripts for drama rehearsals, creating role-play chatbots of literary characters, and supporting translation for English learners—that can free teachers from administrative tasks and allow more time for instruction. He gave a specific example of a teacher who "in 15 minutes" used AI to customize a student's day after IEP changes.

At the same time Shulkin warned of concrete risks. He cited a medical-model example in which an algorithm began to associate a clinic logo with higher cancer probability, an outcome researchers later traced to biased training data. He also raised privacy and identity-theft concerns for children, environmental costs from increased data-center use, and potential disruption to entry-level jobs.

Shulkin described a district platform called Magic School AI and said approved models "are not taught by any data we provide it" and that student interactions "are not training anybody else's models. It stays within us." He urged the district to build AI literacy—skills to evaluate outputs, identify bias and hallucinations, and use AI ethically—into curriculum and professional learning.

During Q&A, Shulkin said teachers already detect anomalous student work by comparing writing histories and that initial responses generally begin with teacher conversations and document review rather than automatic penalties. He said the district is developing toolkits and guidance so teachers can respond consistently.

No formal policy vote on AI occurred at the meeting. Shulkin and other staff said the AI steering committee and work groups will continue developing engagement, metrics and alignment in the coming school years.

Provenance: presentation and Q&A at the Oct. 27 board meeting (see transcript excerpts).