Tech

Face recognition and risk assessment: 10 days in jail for algorithms

[ad_1]

In the USA, the third case of a false arrest based on facial recognition software has become known; this time a second algorithm ensured that the person concerned had to stay in prison for ten days. The New York Times reports on the 33-year-old’s case and has since filed a lawsuit. His case should confirm critics who object to the fact that the police and judiciary in the USA increasingly rely on algorithms, although their inadequacies are known. That the supposedly neutral technology can also be racist was discussed more and more this year.

The case that has now become known therefore occurred in February 2019: After two police officers in the US state of New Jersey had escaped an alleged – African-American – shoplifter, they had sent the photo from his fake driver’s license to investigators with access to facial recognition software. The then issued the uninvolved 33-year-old African American as a hit. Although he has now been able to prove that he had made a transfer elsewhere at the time the wanted man had fled, he had to be behind bars for ten days. An algorithm for determining the deposit had rated the risk as too high to be released. So he was in jail for two algorithms.

The criticism of racist algorithms is not new. In 2016, US journalists reported that technology for the automatic determination of the risk of recidivism of offenders systematically disadvantages African Americans. In the case of the now arrested 33-year-old, the decision was probably based on two previous arrests for drug possession. According to the report He has therefore now even considered to plead guilty despite his innocence in order to avoid an automatic higher penalty for the third offense. Only the proof that he was more than 40 kilometers away at the time of the crime prevented that.

The debate about discrimination through algorithms is likely to gain momentum with the new case: All three people whose false arrests based on facial recognition have become known were black, according to the New York Times.


(mho)

To home page

.

[ad_2]