Best of the Week

Latest

AP shows how AI tool for adoption matchmaking fails foster kids

AP23166757731897_HM.jpg

Through dogged reporting and the extensive use of public records, Sally Ho and Garance Burke uncovered how an artificial intelligence-powered tool has fallen short of its claim to be a technological revolution for the world of child welfare. The reporters interviewed more than two dozen social workers, parents and AI experts to understand the complexities of adoptions involving vulnerable children in the foster care system.

They filed two dozen local, state and federal public records requests and aggressively fought for access until they received hundreds of responsive documents. That trove allowed the reporters to shed light on how the proprietary Family-Match algorithm has scored children and families using more than 250 sensitive data points — including whether foster youth were sexually abused or have a criminal record — something many agency officials say they did not even know.

Contact us