Post

1/ I suspect we are days away from hearing Elon is "using AI" to “streamline” benefits programs. You should know why fully automating decisions about who gets critical assistance is a terrible idea that will have disastrous consequences. A 🧵 based on work I’ve done with the great @awhf.bsky.social

2/ AI tools can help long-understaffed and underfunded agencies quickly process lots of data and respond with greater speed to the needs of constituents. BUT careless & punitive design and implementation of these tools have often prevented benefits from reaching the people they’re intended to help.

3/ In 2016, during an attempt to automate eligibility processes, Indiana denied *a million* public assistance apps in 3 years—a 54% rise. Those who lost benefits: a 6 y/o with cerebral palsy and a woman who missed an appointment with her case worker b/c she was in the hospital with terminal cancer.

4/ That same year, an Arkansas assessment algorithm cut thousands of Medicaid-funded home health care hours from people with disabilities.

5/ And in Michigan, a new automated system for detecting fraud in unemployment insurance claims identified fivefold more fraud compared to the older system—causing some 40,000 people to be wrongly accused of unemployment insurance fraud.

6/ These programs aren't perfect, and oversight is critical. But these cases show how tech systems keep qualified people away from help. They are the modern face of an old problem: The U.S. has long put up administrative burdens to deter the poorest applicants from making use of government help.

7/ The New Deal–era distinction between “earned” social benefits—like Social Security—aimed at working white men, and social assistance programs for the poor has translated over time to especially high barriers for the Americans most in need. AI now turbocharges this restrictive approach.