In 2016, Microsoft released a chatbot named “Tay” on Twitter as an experiment in conversational learning. The experiment quickly ran into problems as Twitter users started tweeting at the bot with misogynistic and racist comments, and “Tay” started repeating that sentiment back to users. More recently, Google released a predictive text feature within Gmail where the algorithm made biased assumptions referring to a nurse with female pronouns. Google eventually stopped the feature from suggesting pronouns.
Firth-Butterfield also pointed to the fact that our most popular virtual assistants from Apple, Amazon and Google have female voices and are designed to serve and take orders.
Programmers “are replicating the fact that people who take calls in call centers tend to be women but we could create a different world with AI if we had more diverse teams creating [it],” she said.
Not all the benefits of diversity are immediately obvious, but the report says that changing the ratio will require proactive measures.
“When you bring more people into the field you get more creative outcomes,” said Tess Posner, CEO of AI4All, a non-profit working to increase diversity in fields including artificial intelligence.
Firth-Butterfield and Posner both argue that companies with greater diversity perform better.
“We see that diversity improves innovation, and the technology itself,” Posner said. If people from a variety of backgrounds are building AI systems, “it will better reflect society,” she said.
Be the first to comment