[ad_1]
The bill would apply only to companies that earn more than $ 50 million a year or who have data for at least a million people or devices. Small businesses would be theoretically safe.
Senators considered that it was a question of civil rights and cited examples of recent incidents. Facebook is still accused of housing discrimination after allowing advertisers to exclude people in a racist or badist manner, while Amazon has closed an automated recruitment tool after being deemed discriminatory against women. . Facial recognition also has bias issues. "This is a modern form of practices like" real estate steering "(black couples have been deterred from settling in some neighborhoods), said Senator Booker, but more insidious in that he is "much more difficult to detect. "In theory, this would prevent companies from ignoring the risk of bias.
There is no guarantee that the bill will become law, and there are questions about its effectiveness. Would a company always be held responsible if a biased algorithm went through the cracks? What happens if an algorithm does not cover potentially biased data? As useful as legislation may be to ensure equity, it may also create uncertainty and additional overhead costs that may not always be necessary.
Source link
