A Michigan government agency wrongly accused individuals in at least 20,000 cases of fraudulently seeking unemployment payments, according to a review by the state.
The review released this week found that an automated system had erroneously accused claimants in 93% of cases – a rate that stunned even lawyers suing the state over the computer system and faulty fraud claims.
"It's literally balancing the books on the backs of Michigan's poorest and jobless," attorney David Blanchard, who is pursuing a class action in federal court on behalf of several claimants, told the Guardian on Friday.
The Michigan unemployment insurance agency (UIA) reviewed 22,427 cases in which an automated computer system determined a claimant had committed insurance fraud, after federal officials, including the Michigan congressman Sander Levin, raised concerns with the system.
When we give up human judgment in favor of software, code becomes law and programmers our unelected legislators.
(Score: 1, Interesting) by Anonymous Coward on Monday December 19 2016, @02:06PM
> 93% failure rate is *worse* than chance, FFS; a coin toss would generate more reliable results.
No its not. Or at least the stat on its own is not enough to say.
That's because it doesn't say anything about the total sample size, just the ones that were flagged for fraud.
So there could have been a million claims examined and only ~23,000 were flagged as fraudulent. That would still leave 977,000 not flagged for fraud.