“AI could widen the gaps we've spent decades closing.”

Both futures are real. AI can put a patient tutor beside every struggling reader, translate for every family, and differentiate materials at a scale no teacher has time for. The same tools can also deepen digital divides, encode algorithmic bias into screening and discipline patterns, fail students who use assistive technology, and violate civil-rights obligations — often without the district ever knowing, because the harm hides inside a vendor's model.

The difference between the two futures is not intention; it is instrumentation. Equity in AI adoption is a set of auditable practices: who has access to which tools (and on which devices and bandwidth), how tools are screened for bias and accessibility before purchase, how outcomes are monitored by student group afterward, and whether multilingual learners and students with IEPs were designed for or bolted on.

Concretely: accessibility review against WCAG and assistive-technology compatibility; a bias and disparate-impact review protocol; procurement questions with auto-disqualifiers; and family communication that reaches every language your community speaks.