Loading...
Loading...
Wisconsin's education AI market is defined by a geographic split as stark as any in the Midwest: the Milwaukee metro, where Milwaukee Public Schools serves 65,000 students with some of the deepest poverty concentrations in the Great Lakes region, and the Madison corridor anchored by the University of Wisconsin-Madison and Madison Metropolitan School District, where the state's highest concentration of education researchers sit a 90-minute drive from school systems whose implementation capacity is among the nation's lowest. The Wisconsin Department of Public Instruction (DPI) under State Superintendent Jill Underly has been navigating this divide while developing AI in education guidance — with an explicit equity lens that acknowledges the MPS-Madison performance gap and the role technology can play in exacerbating or reducing it. The University of Wisconsin-Madison is the state's flagship research institution and produces foundational education research across cognitive science, data analytics, and learning technology. UW-Milwaukee, 15 minutes from MPS headquarters, is the more directly practice-oriented research partner for Milwaukee urban education. Epic Systems in Verona — the nation's largest electronic health records company with 12,000 employees on its campus near Madison — intersects with education AI in a specific and underappreciated way: Epic's patient data infrastructure, which includes pediatric health and learning-related condition data, has been the subject of research partnerships with UW-Madison that have produced ML models for identifying students with undiagnosed learning disabilities earlier than standard diagnostic pathways.
Milwaukee Public Schools is simultaneously the state's most urgent AI education opportunity and its most complex deployment context. With 65,000 students — 80% qualifying for free or reduced lunch, 15% English language learners serving a large Hmong, Spanish, Somali, and Burmese community — MPS has the scale to negotiate meaningful AI tool contracts and the demographic diversity to generate significant outcome research. The district has been using MTSS (Multi-Tiered System of Supports) data analytics tools since 2020, and the MPS Office of Accountability is one of the more data-capable urban district offices in the Midwest. However, MPS's technology infrastructure varies enormously across its 160 schools — some operated directly by the district, some through its partnership-school model with Milwaukee College Prep, Ronald McDonald House charities-affiliated schools, and independent charter operators — creating a fragmented technology landscape that makes district-wide AI deployment significantly more complex than a unified district. The pattern we've observed in Milwaukee: AI tools that require consistent district-wide data infrastructure tend to underperform their pilots because the pilot schools are typically the ones with better technology, skewing results. AI tools that work with variable data quality — adaptive platforms that function on minimal historical data and build their models from student interactions forward — have shown better scalability across MPS's heterogeneous school context. MPS's 2024 adoption of AI-powered multilingual parent communication tools, specifically for its Hmong and Spanish-speaking communities, has been the highest-engagement AI investment the district has made — parent response rates to AI-assisted attendance notifications in Hmong exceeded the same communications in English.
UW-Madison's Wisconsin Center for Education Research (WCER) is one of the oldest and most productive education research centers in the United States, and it has been increasingly active on AI in education since 2022. WCER's collaboration with Epic Systems on learning-related health data has produced a specific research thread that almost no other institution is pursuing: using Epic's pediatric health records — with appropriate IRB protocols and de-identification — to identify early patterns in vision, hearing, and developmental health records that predict reading difficulty before kindergarten diagnostic assessments would flag it. The research is genuinely novel and the implications for Wisconsin's early literacy AI market are significant — a screening model that uses birth-through-age-5 health data to predict K-2 reading intervention need could reshape how districts allocate AI reading tools before the first standardized assessment. UW-Milwaukee's School of Education has a more direct Milwaukee urban education focus, with research partnerships with MPS and the Milwaukee-area suburban districts including Wauwatosa, West Allis, and Racine Unified. UWM's 2023 launch of an AI-readiness assessment tool for Wisconsin school districts — a free diagnostic that helps district leadership understand their data infrastructure gaps before purchasing AI tools — has been adopted by 60+ Wisconsin districts through the DPI's technology planning process. This tool, developed with DPI support, is the entry point assessment that DPI now recommends all Wisconsin districts complete before AI tool procurement.
Madison Metropolitan School District at 27,000 students occupies a peculiar position in Wisconsin education: it serves a population that ranges from UW-Madison faculty children to some of the highest concentrations of Black and Hispanic poverty in Dane County, in the same school buildings. MMSD's achievement gap — among the largest racial achievement gaps of any school district in the nation, as documented by WCER research — has made the district an intense focus for equity-driven AI interventions, and also a cautionary case about what AI tools cannot fix without structural changes in how schools serve different student populations. The district's 2022 pilot of AI-powered personalized learning platforms in two middle schools produced results that MMSD's research office reported publicly: measurable gains in mathematics for white and Asian students, no measurable gain for Black students in the same classrooms with the same tools. The analysis attributed the differential to teacher implementation variation — teachers were more likely to facilitate structured AI tool use for students they perceived as engaged, regardless of actual performance need. The research has since influenced DPI's AI guidance, which now requires human-centered implementation audits alongside technology outcome data. Madison Metropolitan's per-pupil spending of $18,000+ gives it resources that most Wisconsin districts don't have, and MMSD has used that capacity to run more rigorous AI tool evaluations than almost any comparable-size district in the state — making it a valuable, if complicated, evidence base for Wisconsin AI adoption decisions. Per-seat costs for the adaptive learning platforms MMSD has deployed run $22–$40 per student annually, with implementation and research evaluation adding $75K–$150K per year district-wide.
Training teams on AI tools, managing organizational change for AI adoption
Building conversational AI for customer service, sales, and internal use
Predictive models, data analysis, and ML pipeline development
Bespoke AI solutions, model fine-tuning, and custom model development
UW-Madison's Wisconsin Center for Education Research has been partnering with Epic Systems to study whether pediatric health records — birth through age 5 — can predict reading difficulty before kindergarten assessments identify it. With appropriate IRB protocols, de-identified Epic pediatric data has been used to develop ML models that flag developmental patterns predictive of K-2 reading intervention need. The research is in academic publication stage and not yet in commercial deployment, but several Wisconsin districts are in early conversations with WCER about implementing screening protocols built on this model.
MPS has deployed AI-powered multilingual parent communication tools that have shown strong engagement among Hmong and Spanish-speaking families, with parent response rates for Hmong-language attendance notifications exceeding English-language rates. The district also uses MTSS data analytics for academic intervention tracking. The primary AI deployment challenge in MPS is infrastructure variance — the district's 160 schools span directly-operated and partnership-model schools with different technology environments, making district-wide AI rollouts that depend on consistent data infrastructure harder to execute than pilot results suggest.
Madison Metropolitan School District's 2022 AI pilot in two middle schools found measurable mathematics gains for white and Asian students but no measurable gain for Black students using the same tools in the same classrooms. MMSD's research office attributed the differential to teacher implementation variation — teachers more consistently facilitated AI tool use for students they perceived as engaged, creating implementation bias that the technology couldn't overcome. DPI incorporated the finding into Wisconsin's AI guidance, adding a human-centered implementation audit requirement to district AI deployments.
Wisconsin DPI's AI in Education guidance, developed with input from UW-Madison and UW-Milwaukee, recommends that districts complete an AI-readiness assessment before procurement. UWM's School of Education developed a free diagnostic tool that evaluates district data infrastructure, staff capacity, and equity framework readiness — adopted by 60+ Wisconsin districts since its 2023 launch. DPI now includes the UWM readiness assessment as a recommended step in its technology planning process, making it the practical entry point for AI vendor conversations.
For Milwaukee Public Schools' scale, enterprise adaptive learning contracts run $15–$28 per student annually, with MPS's partnership-school model creating additional procurement complexity. Madison Metropolitan, at $18,000+ per-pupil spending, pays $22–$40 per student for the more sophisticated platforms it deploys. Smaller Wisconsin districts — which comprise the majority of the state's 421 districts — typically access AI tools through DPI consortium contracts at $12–$22 per student. Wisconsin's per-pupil spending ranks in the middle nationally, but the concentration of resources in Madison-area and suburban Milwaukee districts means rural Wisconsin and mid-size industrial cities like Racine and Kenosha operate with significantly tighter technology budgets.