Democrats Present Bill that Requires Tech Firm to Check for Bias in Algorithms

AI

Senator Ron Wyden and others are now making laws to check the bias in algorithms in social media sites like Facebook and Twitter / Photo by: JD Lasica via Flickr

 

American lawmakers presented legislation on Wednesday that mandates major tech companies to audit machine learning systems for bias in a bid to construct transparency and accountability into the emerging generation of world-changing technology.

Tech news site Gizmodo reports that Democratic Senators Ron Wyden and Cory Booker have introduced the Algorithmic Accountability Act, along with Democratic Congresswoman Yvette Clarke, who presented a similar bill in the US House of Representatives.

It also states that artificial intelligence and machine learning currently operates a seemingly extensive sweep of key processes and tools, such as facial recognition, automated vehicles, ad targeting, customer service, moderation of content, policing, hiring, and even dispute.

While it's interesting to see how such uses vary from one another, it's still a mystery how those decisions were made and if they are truly fair or not. That issue prompted lawmakers to this bid to open the "black box."

The new legislation would have the Federal Trade Commission develop regulations that require tech firms to conduct "impact assessments" of automated decision systems to study the decision-making systems and training data "for impacts on accuracy, fairness, bias, discrimination, privacy, and security.”

Gizmodo says the bill targets tech companies that generate more than $50 million annually or stores data of more than a million individuals.

Meanwhile, Wyden said computers are becoming more involved in many key decisions that citizens make in respect to their everyday lives—be it their capacity to buy a home, get hired, or go to jail.

Such a scenario was seen last year at Amazon, in which the AI leader used the technology to determine the applicants it should interview for available jobs. Following two years of the design launch, they found that female applicants were moved to the back of the list.

AI Now Institute’s Kate Crawford said this incident indicates that it's a lot more difficult to automate these tools than believed and that Amazon's automated application selecting tool "tells you something about the pile of résumés that they had.

"What were they training it on? What was the training data? Surprise, surprise: a lot of white dudes in basically their entire engineering pool.”

But with the Algorithmic Accountability Act, modern technology tools will be ensured to be used to empower people instead of marginalizing them—all while "also bolstering the security and privacy of all consumers,” Clarke said.