Sexy Problems
There are a lot of problems in the world. Some are big, some are small, some are solvable, and some aren't. In this post, I want to talk about a specific class of problems - ones I call "sexy problems.”
Sexy problems are those that are fun to talk about. They're the perfect problems for armchair generals, Monday morning quarterbacks, and backseat drivers. No one is going to solve them so everyone gets to have their opinion. They're often somewhat abstract or futuristic so everyone can have an opinion without lots of real-world consequences, although they have the potential for consequences. Let’s look at some examples.
Deep Fakes
Deep fakes are my favorite example of a sexy problem. Deep fakes are the use of deep learning technology to create realistic videos or pictures. It's basically like advanced photoshopping but now works for video and audio as well. To be clear, sexy problems are real problems that are worth carefully thinking through. Deep fakes have the potential to cause real harm to society. But the amount of exposure they get is wildly disproportionate to their significance.
People are worried that deep fake technology will be used to create fake videos that mislead people. This is especially the case in politics, where the market for this type of thing seems high. Bullshit news has been a problem for a long time and will continue to be. I think it's a huge problem and have written about it over and over again. But here's the thing, deep fakes are a tiny subset of the bullshit in the media today and will probably continue proportionally. Remember that video of Nancy Pelosi's slurred speech? That was just a slowed-down video - it wasn't a deep fake, but the media took it as an opportunity to talk about deep fakes because they're so sexy.
I really don't think deep fakes are going to be as big a problem as people say. I'm sure people will read this and say I'm so wrong, so let's compare deep fakes to a different type of problem, one that is decidedly NOT a sexy problem.
According to the CDC, there were 47,600 opioid-involved deaths in 2017. That's over 100 deaths a day. I contend that there will be more deaths directly related to opioids today than will be from deep fakes over the next ten years combined. And I don't think, upon reflection, that many people would seriously disagree with me.
Deep fakes are worth talking about. But we should acknowledge when we go back and forth about the potential impact of people being misled at some future date, that we're talking about a subject because it's sexy. If we wanted to talk about a more serious subject, we would talk about why over a hundred people are likely to die today because of opioids.
Self-driving Cars and the Trolley Problem
Another example is the trolley problem, which asks "Would you redirect a runaway trolley to kill one innocent person when it would otherwise kill five?" This is an interesting thing to think about, and especially with the rise of self-driving cars, there's increased interest. What should a self-driving car do when choosing between killing one or killing many? Should we even let self-driving cars on the road until we have an answer?
Fun questions to think about, fun to come up with frameworks for why one action is better than the other, and fun to debate with friends. But I have one minor quibble with it: I've never faced a “kill many people through inaction or one through action problem” while driving. I have no idea what I would do in this situation. Would I swerve away from a large group to hit a smaller number? I have no idea. It is an interesting philosophical question, but no one tested me on it before giving me a license, so I don't see why we should test self-driving cars on it. It’s interesting to consider, but let’s modulate our attention to sexy problems by making sure we focus on the most important ones - sexy or not.