Stories

Civic Science Fellow Michael Akinwumi Develops Tools for Responsible AI Governance

As artificial intelligence reshapes our world, the need for informed governance is increasingly urgent. Yet at the state level, where many crucial decisions about AI implementation will be made, some states are far more ready for this challenge than others. Michael Akinwumi is working to bring more of them up to speed.

“At the state level, we only have about 4 percent of lawmakers who have science, engineering and healthcare backgrounds,” says Akinwumi, a Civic Science Fellow at Rutgers University’s Eagleton Institute of Politics. “Right now, AI itself is a scientific endeavor, an engineering endeavor. So the question is, who wants to fill the 96 percent knowledge gap here?”

Akinwumi, who also serves as Chief AI Officer at the National Fair Housing Alliance, brings a global perspective to this challenge. With a background spanning Nigeria, Canada, and Europe before landing in the United States, he’s seen firsthand in many venues how AI can both expand opportunities and exacerbate inequalities.

The AI Revolution: Promise and Peril

Credit scores offer an instructive example of AI’s potential to either improve or worsen a process that many people rely on. “Before the advent of AI systems, the technology behind credit scores was so weak that it could not leverage newer information, what we call alternative data,” Akinwumi says. This meant that people without traditional credit histories, like those who haven’t used banks or just moved to the United States, can struggle to qualify for mortgages, open bank accounts, or get credit cards. Akinwumi has experienced this personally. “I was once a new immigrant, and I had to wait three, four years before I could even have a credit report.”

AI-based systems have the potential to change that by analyzing alternative data sources such as rental and utility payments to create more inclusive credit profiles. But without proper oversight and implementation, these same systems could reinforce existing biases—or create new ones.

Similar dynamics play out in job applications, where AI-powered tools are increasingly common. “Job seekers, especially those who have access to premium versions of some of the generative AI systems, can use them to draft resumes that are more likely to be considered for the next round of the application, even when the reviewer is a person,” Akinwumi says. While this could widen the divide between those with access to advanced AI tools and those without, the technology also has the potential to help level the playing field, making it easier for less capable writers to make themselves shine on a resume.

Scenarios like these underscore the urgent need for informed AI governance, including at the state level. Yet currently, only a handful of states—notably Utah—have the kind of “regulatory sandbox” that allows for collaboration and information sharing among representatives from government, academia, industry, and civil society at large.

Akinwumi’s fellowship project aims to change that. He is building an AI-powered tool to help state legislators and citizens navigate the complex landscape of AI risks and opportunities. The tool will allow users to select their state and the sector they’re focusing on—currently available is healthcare, housing, or finance, with more to come—and receive tailored policy recommendations.

“It will be a website or application where state lawmakers, and especially people working with state lawmakers like an office of legislative services, business leaders, and nonprofit staff can go,” Akinwumi says. “State lawmakers can use this to not only mitigate the risk, but also make sure that the dividends of these systems are equitably or equally accessible to everyone.”

The app will scour the state records for current bills touching on AI and then aggregate and analyze them. Finally, it will offer a list of things to consider for inclusion in the bills to ensure the AI application fair and safe.

For example, the app might recommend that regulators make sure that AI systems managing affordable housing prevent profit motives from overriding social responsibility. It then can break that recommendation down into specific actions, such as drafting clear regulations, establishing penalties for wrongdoing, and creating a public reporting system.

People using the tool will be able to ask the system to explain its reasoning more, push back or make alternative suggestions, and also potentially add information for the tool to aggregate and analyze. “I imagine that people using it are coming from different political backgrounds, so their priorities will be different,” he says.

A Range of Readiness

While planning this tool, Akinwumi quickly realized that a one-size-fits-all approach wouldn’t work. “I was shocked when I started looking for bills at the state level. Because the U.S. is so technologically advanced, I thought all the states would be at the same level as the federal,” he says. Instead, he notes, “if we make these policy recommendations, some states may not be ready to implement them because of where they are in terms of their AI maturity.”

This realization led to the development of an additional part of his tool: an AI Readiness Index.

The index considers factors like whether the state has an AI task force, workforce development initiatives, data infrastructure, or a responsible AI framework to create a state’s readiness score. Using the score, the tool can tailor a state’s policy recommendations to what is actually feasible for the state right now.

This individualized approach reflects Akinwumi’s broader philosophy on AI governance. “What we need to focus on is really how AI is going to be used, not just the technology itself,” he says. “The specific applications of AI are really what is going to be consequential to most people.” This means moving beyond blanket regulations to consider the impacts of AI in different sectors and communities.

Filling the Expertise Gap

Akinwumi’s work highlights a critical knowledge gap in the current landscape of AI governance. At the federal level, agencies like the Office of Science and Technology Policy (OSTP) offer relevant expertise and guidance on emerging technologies. But at the state level, where many decisions about AI implementation will ultimately be made, such resources are scarce.

“We don’t have the equivalent of OSTP at the state level,” Akinwumi says. Even Utah, the state that’s furthest ahead, has only just created a similar office but hasn’t staffed it yet. Akinwumi sees this as a chance for both policymakers and interested funders to have a hand in shaping the future of AI governance.

“There’s an opportunity here for funders to make sure that every state and territory has a policy office where civil servants, scientists, engineers, and other stakeholders can take part in the conversation,” Akinwumi says. Such offices could also serve as testing grounds for AI applications, ensuring that technological advancements truly serve the needs of all the state’s communities.

From Development to Deployment

As his fellowship progresses, Akinwumi is already thinking about the long-term impact of his work. “I want to make sure that this tool is adopted by lawmakers, and even by common citizens like myself, so that they can have conversations with elected lawmakers about how AI is being used and how to protect beneficial uses.”

Akinwumi hopes to have the app completed by June, at which point he will pivot to engagement. He plans to travel to different states to meet with people who are working on some of these bills and show them how the tool works so they can test it and hopefully adopt it.

This focus on practical implementation and community engagement is at the heart of Akinwumi’s approach to civic science. He emphasizes that shaping responsible AI policy isn’t just a task for technologists or policymakers. “It is really a collective responsibility to make sure that AI is used responsibly. And that doesn’t require having a PhD in computer science or math,” Akinwumi says. “It’s really being familiar with how this technology can be used in our current lines of work and how to make sure that it’s used in a responsible way.”

As AI continues to transform much of our lives, from healthcare to housing to employment, Akinwumi’s work calls attention to the pivotal role of civic science in ensuring these changes benefit all members of society.

By bridging the gap between knowledge of cutting-edge technology and state-level policymaking, he’s helping to create a future where AI serves as a tool for equity and opportunity, rather than a force that exacerbates existing divides. One state at a time.

Michael’s Civic Science Fellowship at the Eagleton Institute of Politics is supported by the Rita Allen Foundation.