AI Brings Both Risks and Benefits for People With Disabilities

On June 10, the United Nations hosted a conference titled Artificial Intelligence for Inclusion: Strengthening Workforce Participation for Persons with Disabilities. Organized by the Permanent Mission of Canada to the UN, the event brought together a panel of experts from multiple sectors to explore how AI tools can foster inclusive workplaces that promote fairness and accessibility.

Since the rise of generative AI systems in the early 2020s, many industries have been reshaped. For workers across the globe, AI has streamlined routine tasks and boosted efficiency, enabling many to move into higher-level roles. The result: new jobs, greater productivity, and new opportunities.

However, these benefits have also come with significant risks. AI systems can introduce unintentional bias, particularly in hiring processes, undermining diversity and merit-based employment. Many systems are designed with able-bodied users in mind, excluding people with disabilities and reinforcing existing inequalities.

During the conference, panelists addressed how AI can be designed to better support people with disabilities—who have long faced disproportionate barriers to employment and inclusion. Because AI is still relatively new in global workplaces, many industries lack the necessary safeguards to prevent discriminatory outcomes.

“AI is transforming the way we live, not just how we do business,” said Patty Hajdu, Canada’s Minister of Employment and Workforce Development. “Its rapid arrival and lack of regulation pose major questions around inclusion, ethics, privacy, and our core institutions.”

Dr. Jutta Treviranus, Director of the Inclusive Design Research Centre at OCAD University, warned that many AI systems rely on algorithms that discriminate against anyone who falls outside statistical norms. Roughly 90% of U.S. organizations, she noted, now use AI in hiring or disciplinary processes—often excluding those considered “outliers,” including many disabled individuals.

“Bias toward optimal patterns means bias against difference,” said Treviranus. “As AI improves, it becomes better at discrimination. Programs designed for efficiency often exclude anyone who isn’t 'optimal.'”

She also criticized current risk assessments in the U.S., which consider outcomes for statistical minorities as “insignificant,” reinforcing what she calls "statistical discrimination."

AI tools that do aim to support people with disabilities often focus only on physical impairments, overlooking intellectual and cognitive disabilities. Disabled women, in particular, are at high risk of being excluded due to overlapping biases in data.

“AI can be a powerful equalizer—but only if developed with intentionality,” said A.H. Monjurul Kabir, Senior Global Adviser on Gender Equality and Disability Inclusion at UN Women. “It must not deepen existing stigma, discrimination, and inequality—especially for women and girls who face compounded barriers.”

Treviranus emphasized that even proportional representation in data sets won’t solve everything: “If you are highly unique, privacy protections can actually make things worse. Differential privacy removes the very data that would help make systems more inclusive for you.”

Access to AI-powered assistive technologies remains limited for many disabled individuals around the world. In some regions, a lack of electricity alone can prevent people from using basic breathing aids or digital tools.

“People with disabilities often pay more for technology that works poorly or breaks down easily,” Treviranus said. “The farther you are from the average user, the worse the tech works for you.”

There are some hopeful developments. In Germany, hotel workers with intellectual disabilities are being provided with tablets to help manage tasks. In Sri Lanka, the FAO’s CIRCULAR project—funded by the EU—is helping reduce single-use packaging and improve accessibility in food retail. Meanwhile, Kenya and Uruguay are reforming agricultural policies to reduce chemical use and environmental harm, supported by the FAO's FARM project.

Jürgen Dusel, Germany’s Commissioner for Persons with Disabilities, stressed the need for broader reform. “We must ensure AI supports the most underserved communities,” he said.

Treviranus concluded by highlighting the value of designing AI systems with the most marginalized people in mind: “If we focus on those facing the greatest barriers, we can build more adaptive, efficient, and inclusive systems. It’s not just the right thing to do—it’s the smartest thing to do.”