Spike Jonze’s new film Her has sparked some fascinating conversations about artificial intelligence, gender and how we might traverse the uncanny valley we experience when real worlds and virtual worlds overlap.
Liat Clark from Wired contacted me for a great piece about AI virtual assistants and he asked me why it was that Hollywood usually depicts friendly AI as female and threatening AI as male. I think that, even though we may be more aware of gender stereotypes and how limiting and self-destructive they can be, it doesn’t mean that we don’t invoke them in the stories that we tell about ourselves. We humans are caught in some very deep cultural grooves: we still tend to associate a helpful, nurturing creature with feminine characteristics and an aggressive and destructive force with male characteristics. To reverse that association would only result in reinvoking it: everyone would notice that it was “backwards.” In that respect, storytellers cannot help but invoke the tropes that define us – particularly when they’re weaving tales about artificial versions of ourselves.
Clark was also interested in why Hollywood tropes about love and romance always seem a bit retro, lagging behind the current zeitgeist. I think that the representations that define our popular culture are profoundly disconnected from reality and are more likely to reflect a marketer’s hunch about what a particular demographic craves rather than what real living people actually want. One reason I’m so excited about the prospects for social media is that it gives marketers and media companies better information than they’ve ever had before about what animates us, what preoccupies us, and what we care to share (this is the topic of one of my TED talks). I think marketers, advertisers, programmers and creators of entertainment content will need to respond to increasing pressure to supply what audiences actually want – rather than producing hackneyed stories based on primitive stereotypes.
Clark turned out to be much more skeptical than I was about just how far we can go with AI. As my friends well know, I am borderline obsessed with the possibilities of the singularity, and so I’m convinced that we’ll be able to make life-like versions of ourselves in my lifetime. For better or worse, I believe that we’ll make AI that conforms to our current notions of perfection, whatever those happen to be at the time. I also think we will undoubtedly change our definition of perfection the moment we think we’ve achieved it. We’re sort of predictable that way.
The broader question may be why we are so obsessed with humanizing technology in the first place. There’s a great scene in Prometheus in which the robot Michael (played brilliantly by Michael Fassbender) puts on some protective head gear which is entirely unnecessary for him. When a human asks him why he bothers, he says it’s because humans are far more comfortable with creatures that act and look like them. Bingo. Yes, we quite capable of empathy, but we are also deeply self-involved creatures, hard-wired for self-preservation, and anthropomorphism is a crutch we’ve been using for millennia. Our virtual assistants and the robots we engineer will no doubt reflect our knee-jerk discomfort with anything elementally different from ourselves.