Opinion

A key part of any solution should involve revolutionizing data collection systems. We often don't have the data we need to build unbiased AI, whether it is the lack of uniformity in electronic health records, state-by-state variation in criminal sentence reporting, or missing data in the 2020 Census. The solution will require increased funding to agencies that gather data in underrepresented communities, outreach to community organizations to gain local buy-in, and democratization of data so that citizen scientists can highlight disparities and identify ongoing gaps.