A law mandating regular audits of bias for programs that mechanically evaluate employees with artificial intelligence (AI) goes into effect for the primary time in Latest York City.

Bloomberg News reported that Latest York City, the U.S., has begun full-scale implementation of the local corporation’Automated Employment Decision Tool (AEDT) Act’ on the fifth (local time).

Here, AEDT refers to an AI program used for evaluation for hiring, continuing employment, and promotion of employees.

The law was enacted by the Latest York City Council last 12 months and went into effect in January this 12 months. Nevertheless, Latest York City didn’t have specific guidelines related to AI audits, so implementation was delayed, and after preparing detailed regulatory plans in an issue and answer (Q&A) method, it began to be applied from that day.

In accordance with this, Latest York City businesses must undergo an annual bias audit from an independent auditor for his or her AI tools for human resources and recruitment. Using tools which have not been audited is prohibited.

As well as, the outcomes of an independent audit of the tool should be made public. The general public report includes information concerning the algorithms in use, in addition to ‘average scores’ by different races and ethnicities and gender.

The penalty for violating this law is $375 (about 490,000 won) for the primary violation, $1,350 (about 1.76 million won) for the second violation, and $1,500 (about 1.96 million won) for 3 or more violations.

The law applies to all staff employed by local businesses, even in the event that they don’t live in Latest York City. Latest York City and Congress officially accepted the difficulty of AI’s ‘risk of bias’ in recruitment or personnel in 2020 and enacted the ‘AEDT Act’, which mandates audit activities for personnel and recruitment AI systems.

In accordance with Bloomberg, the Latest Jersey legislature can also be considering making AI tools available only to employers who’ve undergone a bias audit, like Latest York City. Moreover, Maryland and Illinois have introduced laws banning the usage of facial recognition and video evaluation tools in job interviews without the consent of applicants.

The California Fair Employment and Housing Commission is reviewing a latest mandate banning AI tools and tests that may screen applicants based on characteristics corresponding to race, gender and ethnicity.

Reporter Jeong Byeong-il jbi@aitimes.com