Marginalized Americans Express Skepticism Toward AI

Artificial intelligence (AI) is often hailed as a transformative force, promising to revolutionize industries and simplify daily life. However, a new study from the University of Michigan highlights a stark divide in attitudes toward AI, particularly among marginalized Americans.

The study, which surveyed 742 people from a nationally representative sample, found that gender minorities, women, and disabled individuals—especially those who are neurodivergent or living with mental health conditions—hold significantly more negative views of AI than their majority-group peers.

"AI may be everywhere, but it’s not for everyone—at least not yet," said Oliver Haimson, an assistant professor at the University of Michigan’s School of Information and Digital Studies Institute. "If we continue to ignore the perspectives of marginalized people, we risk building an AI-powered future that deepens inequities rather than reducing them."

Key Findings

Nonbinary and transgender participants reported the most negative views of AI overall.
Women were significantly less likely than men to view AI positively.
Disabled participants, particularly those who are neurodivergent or have mental health conditions, also expressed negative AI attitudes.
Surprisingly, Black participants held more positive views of AI than white participants, suggesting a more nuanced relationship with the technology among people of color.

Real-World Harms

The study highlights several real-world harms that contribute to the negative perceptions of AI:

Facial Recognition Software**: Systems that misgender or misclassify trans and nonbinary people, often while surveilling them.
Predictive Policing Algorithms**: Algorithms that reinforce racial bias and lead to unjust arrests.
Health Care Systems**: AI models that are not designed with disabled people in mind, leading to exclusion and harm.

"These are not abstract concerns," Haimson noted. "People who are wary of AI often have lived experiences with systems that misidentify, exclude, or harm them due to their gender or disability."

Implications

The research challenges the dominant narrative of AI as a neutral or universally beneficial tool. It calls on technologists, companies, and policymakers to consider the perspectives of marginalized groups and ensure that AI development is inclusive and equitable.

"We need to ask who this technology is working for and who is being left behind," Haimson emphasized. "Only by addressing these issues can we build an AI future that truly benefits everyone."

The study, co-authored by Samuel Reiji Mayworm, Alexis Shore Ingber, and Nazanin Andalibi, was presented at the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT '25), a leading venue for work at the intersection of AI and ethics.