Ethics panels reject and delay biometrics, AI projects for Google, IBM, Microsoft
Since the beginning of 2020, Google has turned down a project to use artificial intelligence to help a client decide whom to lend to; Microsoft has placed restrictions on voice mimicry software and IBM has rejected a client request for an advanced facial recognition system. The reason for all these decisions and more is the introduction of panels of executives and other leaders, reports Reuters after interviewing AI ethics chiefs at each company.
The report finds that the companies are waiting for regulation from governments and blocs, and also to improve their own technologies to avoid issues such as perpetuating biases around race and gender.
Just as the AI sector was exploding in the West, the industry has applied the brakes itself in the interests of social responsibility. Rights activists are demanding that the changes should go even further, with fully independent ethics committees working with full public transparency.
In 2017, Microsoft established its ethics committee as scrutiny around artificial intelligence grew. It was followed in 2018 by both Google and IBM.
As well as the credit scoring rejection, Google has also blocked new features analyzing facial emotions out of fear of cultural insensitivities and has placed under review a Cloud facial analysis service begun in 2015 to categorize photos of people into joy, sorrow, anger and surprise. Emotion recognition is a highly-contentious area of study.
IBM had to balance the development of voice mimicry to build voice services for impaired people with the risk of the software being used for deepfakes. Its AI Ethics Board has begun assessing how to manage the development of implants and wearables wired into the brain, due to the prospect of hackers manipulating the brain, IBM’s privacy officer told Reuters.
IBM also declined a client request not to customize facial recognition technology to detect fevers and face coverings. Its AI board made up of about 20 departmental leaders decided manual checks on people would be sufficient and prove less intrusive for privacy no photos would be retained. Six months later, IBM announced it was discontinuing its face biometrics service.
The companies told Reuters that they welcome clear regulation on the use of AI, that acting responsibly was also in their own financial interest, but that they hope any regulation would be flexible enough to allow innovation. Instagram recently updated its age restrictions for child safety ahead of the UK’s legal requirements for child-centric design for non-verified accounts. It has been joined by Twitter, TikTok and Facebook which are making the changes globally, seemingly because of the new UK requirements for age appropriate design, reports The Guardian.