Chances are, if you’ve got a smartphone, you’re employing a female virtual personal assistant. And if you were involved in developing the AI that powers those assistants, you’ve probably felt the heat from critics incensed about needless adherence to tired gender roles. Apple’s Siri, Microsoft’s Cortana, Amazon’s Alexa, and pretty much every American GPS system has a default female voice. We intuitively understand why this is because we’ve got eyes and ears and maybe a softcover of Friedan, but it’s less clear why more people don’t recast this bit of AI role play to better suit themselves. The interfaces we choose, after all, should suit us, not their creators.
The tech world has been called out on issues of diversity before and, to its credit, has done a fair job responding. Apple, for one, was slapped last year by MTV Act contributor Joey Parker for lacking racial diversity in its emoji. The company responded by giving users the option to change certain emoticons from the newly default yellow color to a variety of different skin tones reflecting different races. If Twitter and Facebook are any indication, emoji users have wholeheartedly embraced this departure from the default. It was a good idea that came from outside the building, but was embraced by corporate, which seems like a solid result.
Emojis are a bit less complicated than voices for both technological and psychological reasons. And that may be why the typical diversification process — person registers complaint, company reacts by creating new product — is running (or failing to run) in reverse.
It was relatively easy for the engineers at SRI International who created Siri to justify her gender using soft science. Data shows that people generally respond better to female voices, but the biases of the larger population are neither a compelling reason to make a decision or justification for homogeneity. It’s easy to make Siri’s voice male and statistically likely that many iPhone users would prefer it that way. Still, default reigns.
The differences in the way humans perceive digital, gendered voices were explored by the late Stanford University communications professor Clifford Nass, whose pioneering studies showed that female voices tended to be perceived as helpful while male voices were authoritative. It makes sense, if we embrace a simplistic understanding of those results, that people would prefer a female voice. But is authority really that bad? Is obsequiousness really what we want from our devices? In a sense, the factory settings are an insult on several levels: Apple doesn’t think you want to be challenged and also thinks that Susan Bennett’s voice isn’t authoritative.
There really shouldn’t be a default setting — smarter engineers and UI people would empower device owners to customize their products — but that decision hasn’t been made yet so we have to empower ourselves. The key isn’t to make a different decision, but to make a personal one. Maybe you prefer a Canadian Woman or a British Man. That’s on you. Siri isn’t a person. Siri is a tool, and the tools we use reflect both our purpose and ourselves.
Yes, personal preferences are subject to cultural biases — the same ones technology firms are so often called out for propagating — but they are not as subject to cultural biases as surveys and market research. We all represent a small sample size, a rounding error, and should feel free to do whatever the hell we want. If that means giving Siri an accent, so be it. He doesn’t care.