Before I was an AI researcher, I was a professional magician. I performed to sold-out theatres in every major Canadian city. On stage, I often used a technique I called “mystified surrender” to keep the audience focused on the experience of the show rather than on trying to solve my tricks. To accomplish this, I devised two opening bits. The first was a mind-reading miracle so impossible-looking that it fooled many seasoned magicians. The second was a card trick wherein I made several mistakes on purpose. I let the observant audience members think they’d seen a crack in the armour. Little did they know, I was setting them up for an even bigger fall. My hope was that by doing these tricks, I could convince them that trying to catch me was frivolous. I was always one step ahead. And thus, through sheer force of being overwhelmed, the audience would cognitively surrender. For the rest of the show, I could take more risks and I could cheat more brazenly, without their prying eyes.

I thought I had left mystified surrender behind when I ventured into academia. But I’ve often found myself surprised by how much the business of AI and magic have in common. My fear is that we have cognitively surrendered to the AI industry, that we have allowed ourselves to be mystified into surrender. AI is a lot like magic; its feats seem impossible. It is a bewildering technological advancement. I worry that we have become so mesmerized by AI, and so overwhelmed by its complexity, that we have deferred our sovereignty to those who created it. That we have deferred the task of imagining what our shared future should look like to technocrats and technologists, who created AI in the first place.

On the surface, this makes sense. AI is an extremely complicated technology. It’s also rapidly evolving. Keeping up with and understanding AI has been my full-time job for the past six years, and even I feel behind the eight ball sometimes. So how can we expect........

© Macleans