It has become a sport among white middle-class men with a penchant for technology to bemoan the failures and inconsistencies of contemporary artificial intelligence solutions (AI). For example, Walt Mossberg recently wrote a much-noted piece at The Verge called “Why Does Siri Seem so Dumb?”
On some level, Mossberg's critique makes perfect sense - it's an honest rant about Siri’s many quirks and problems. For example, Siri is widely believed to be inferior to Amazon’s Alexa because Alexa’s direction microphones (via the Echo Dot, for instance) make her a much more empathetic listener.
But when we read these and other recent discussions of AI more critically, we can’t help but notice a problematic pattern. Mossberg uses a neutral "it" to refer to Siri but still likens his ideal AI to an ideal traditionalist femininity - infallible care, empathy, obedience, and nurture. In this framing, ideal AI seems imagined as some sort of tech bimbo. There is very little surprise then that the absence of these traits warrants an aggressive male intervention. As David Pierce recently wrote in WIRED: "Hey Siri, It's Time to Put Up or Shut Up."
Why do we model AI after our own worst tendencies as human beings? Perhaps it's not AI’s stupidity that's the problem but our own. The definitions of intelligence we adopt, and the biases we necessarily hold, live on in our AI solutions and are then fed back to us as consumers. This generates paradoxes, which have been pretty well studied by consumer researchers such as David Mick and Susan Fournier. Far less attention, however, has been devoted to how these paradoxes are rooted in and also reinforce systemic economic, social, and gender biases.
To explore this further during the holidays, a time of critical reflection and change in the family, we popped an unusual question to both Siri and Alexa: Are you a feminist? Siri apparently holds no views. It simply said: "I'm sorry. I'm afraid I can't answer that." Yet, Alexa had a more interesting comment to make.
Kudos to Alexa's programmers for allowing it to make a political statement that not only serves but also disrupts and challenges some contemporary actors and institutions.
Powerful AI isn’t just about light switches, cooking timers, and Spotify playlists. As it enters more and more domains of our lives, perhaps we’d also love to have virtual companions with interesting views on questions of gender and race, a sense of humor, an ethic, the ability to reason critically, and a self-understanding that can challenge our own.
In the new year, we will start another series of Big Design Lab roundtables, this time focused on "Designing AI Experiences for Growth." One question we will discuss with member CEOs, CMOs, and other AI focused experience designers is what we could gain from adopting a more sociological definition of artificial intelligence, one that actively works with a number of political dimensions and biases as well.
Why does this matter? For one, the AI revolution won’t be worth this label unless it reaches beyond confirming male fantasies of domination and control. And whereas for Mossberg only a fully reliable and unlimited (read: feminized) AI is the “effective weapon” in the coming AI wars, competitively speaking, the best weapon in the AI battle may be a sociological design imagination that can reveal the inherent biases of such a narrow-minded worldview.