'Structural change' necessary to reduce gender bias in AI, say Google, Microsoft leaders

Ensuring that artificial intelligence can be utilized by women with the same success as their male counterparts will require far more than simply updating biased algorithms, according to leaders from Microsoft and Google.

During a panel at the New Rules Summit hosted by The New York Times earlier this month, Peggy Johnson, executive vice president of business development at Microsoft, and Meredith Whittaker, a founder and director of New York University's AI Now Institute and the founder and head of Google's Open Research group, discussed how many tech companies' lack of diversity have led to the creation of distinctly sexist AI.

Because Microsoft's Xbox Kinect gaming system was initially tested only on men aged 18 to 35, it was unable to recognize the movements and commands of women and children. "You want diverse teams because it's good to have diverse teams, but that could have had a real business impact," Ms. Johnson said. "It could have been disastrous."

Ms. Whittaker also noted that many voice recognition tools using conversational AI have trouble recognizing women's higher-pitched voices, and facial recognition has been proven to be far more accurate when used on lighter-skinned men than women and people of color, all due to the datasets used to train the algorithms and the groups tapped to test them.

These issues therefore reflect the need not only for updated, unbiased algorithms, but also for more diverse and equitable hiring practices to begin with. "We need to achieve structural change," Ms. Whittaker said. "The problems are not simply that we're not getting promotions, or opportunity inequity or pay inequity or that women and people of color are more often slotted into contract roles instead of hired full-time, but it is also affecting the world beyond."

More articles about AI:
AI algorithm speeds diagnosis of pediatric gut disease
AI language analysis helps clinicians predict psychosis with 93% accuracy
Community Health Network adopts voice recognition tool to automate clinical documentation

© Copyright ASC COMMUNICATIONS 2019. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months