In this pivotal election year, political maps could decide this fall’s winners and losers. Gerrymandering, the practice of drawing electoral districts to favor specific political parties or demographic groups, remains one of the most powerful yet invisible forces determining whose voices and votes truly count. 

At the University of Michigan, Tyler Simko is fighting back, using artificial intelligence and big data to expose and correct the biases that distort equal representation.

An affiliate of the Center for Political Studies at U-M’s Institute for Social Research, Simko, assistant professor of political science in LSA, uses computation to tackle some of the most persistent challenges in state and local politics. His work includes helping school districts address systemic inequality and building an open-source tool that brings unprecedented transparency to local government decisions.

Simko joined the university in August, after completing his postdoctoral research at Princeton and receiving his Ph.D. from Harvard. His research is informed by a rare personal perspective: Before graduate school, Simko was elected to two terms on the local board of education in his hometown of South Amboy, New Jersey.

“Service in local politics taught me just how much local government decisions shape our lives,” Simko said. “It’s easy to forget in this nationalized era of American politics that most of the decisions that impact us on a daily basis are still made by local governments like city councils and school boards.

“For example, many of the most contentious political issues over the last few years have been primarily state and local decisions like building affordable housing, siting data centers, teaching of Critical Race Theory in public schools, and book bans.”

Mapping fair representation with AI

In his research, Simko uses computational tools and artificial intelligence to restructure political boundaries in more equitable ways. This work includes combating gerrymandering at its source. 

As a co-principal investigator of the Algorithm-Assisted Redistricting Methodology (ALARM) Project, Simko works to pull back the curtain on how these boundaries are created and to provide a more transparent path forward.

The team behind the ALARM project developed open-source software that uses advanced algorithms to simulate the complex, state-specific processes of redistricting, such as the redist software developed by Christopher T. Kenny, Cory McCartan, Ben Fifield, and Kosuke Imai.

The tool, which is free and available to the public, can generate thousands of alternative district maps that strictly follow legal requirements, creating a neutral benchmark. When these thousands of unbiased plans consistently result in more equitable outcomes, they reveal how the maps drawn by politicians are often anomalies that favor partisan interests.

The AI-generated maps can show how states with independent redistricting commissions, like Michigan, draw much fairer maps.

“In recent work, we show that Michigan’s redistricting process is one of the fairest in the country, largely because they have removed the conflict of interest by shifting redistricting power from politicians to an independent commission,” Simko said.

This work has already seen significant real-world impact. The ALARM Project’s tools and evidence are frequently used in litigation, including multiple cases regarding gerrymandering and the Voting Rights Act decided by the U.S. Supreme Court.

Redrawing the lines of school segregation

Simko uses the same computational tools to tackle educational boundaries like school district lines and attendance zones. Having seen firsthand how fragmented districts create vast disparities during his time on the school board in New Jersey, he knows how much those boundaries matter.

“New Jersey has one of the most racially segregated school systems in the country,” Simko said. 

To address this, Simko adapts his gerrymandering detection tools to instead redraw educational boundaries more equitably. His algorithms model thousands of scenarios, balancing priorities like school capacity and travel time while prioritizing goals like program access and socioeconomic diversity.

Crucially, the research demonstrates that districts can often be redesigned to promote integration without requiring new construction or significantly longer student commutes.

By navigating these complex variables, Simko provides school boards with the empirical data needed to address systemic inequality. His work demonstrates that segregation is often a result of political choices rather than geographic constraints, offering a path toward equity rooted in verifiable, data-backed solutions.

An open-source tool for transparency

Simko’s work on making political boundaries more fair underscores another problem: Decisions that profoundly affect people’s lives are often made with little visibility or accountability. Simko’s third major project attacks that issue directly.

While serving on the school board, he realized that neither local leaders nor residents had a clear understanding of what local governments across the country were actually doing.

“The U.S. government is extremely decentralized,” Simko said. “Local governments determine issues like tax rates, school curriculum and land use, but it is very hard to track their policy-making discussions and decisions at scale. There is no centralized place to find information, and no standardized format for the records these governments do share like meeting agendas and minutes.”

To solve this, Simko and his collaborator, Soubhik Barari from NORC at the University of Chicago, used artificial intelligence to co-create LocalView, a massive, open-source database of meeting videos and transcripts from local governments like city councils and school boards across the U.S.

LocalView’s data powers CivicSearch, a tool created by Doug Beeferman at Datamuse, which allows anyone to search discussions in local government on topics like affordable housing, gun violence or sustainability as easily as using a standard search engine. By using computational social science tools to collect and organize these decentralized records, Simko and his collaborators are making the day-to-day work of local governments transparent to the communities they serve.

While researchers use this data to study local governments at a scale never before possible, Simko says this information can also be useful for journalists and members of the public. 

As local newsrooms continue to shrink, Simko sees tools like these filling a critical information gap. 

“This is a place where people can go and learn about what their local government is doing and how it compares to what other local governments are doing,” he said. “There is no way to collect this kind of information at this scale without the use of computational tools and artificial intelligence.

“What I’m most proud of is that all of this data is posted online for free, giving the power of information back to the communities themselves.”

This article is written by Roni Krimgold of U-M Public Affairs.