I don’t usually make predictions, but I’m reasonably confident that my job can be done by a robot in the near future.
It’s not that I’m especially bad at my job, or that my job is easy. It’s just that technology will replace most jobs, and there are already bots that can compose sentences, and I don’t see any reason why this particular innovation will stop short of most journalism.
If you haven’t yet seen the discussion between Daniel Kahneman (Nobel-winning social psychologist) and Yuval Noah Harari (author of Sapiens: A Brief History of Humankind), you should take a look. There’s a lot of big things in there, but here’s the relevant part:
There are a zillion things that the taxi driver can do and the self-driving car cannot. But the problem is that from a purely economic perspective, we don’t need all the zillion things that the taxi driver can do. I only need him to take me from point A to point B as quickly and as cheaply as possible. And this is something a self-driving car can do better, or will be able to do better very quickly.
And when you look at it more and more, for most of the tasks that humans are needed for, what is required is just intelligence, and a very particular type of intelligence, because we are undergoing, for thousands of years, a process of specialization, which makes it easier to replace us. To build a robot that could function effectively as a hunter-gatherer is extremely complex. You need to know so many different things. But to build a self-driving car, or to build a “Watson-bot” that can diagnose disease better than my doctor, this is relatively easy.
And this is where we have to take seriously, the possibility that even though computers will still be far behind humans in many different things, as far as the tasks that the system needs from us are concerned, most of the time computers will be able to do better than us.
My job wasn’t really supposed to be replaceable (at least, in grad school nobody told me it was, and grad school is expensive, so…). But look at this recent quiz from the Times. I’d be willing to bet you won’t be able to tell with total accuracy which passages were written by a bot and which ones by a human — and there’s only eight questions.
It’s a little unfair, yeah: for starters the bot-composed newspaper sentences are formulaic. But computers can do poetry, too. Journalism, the best of it, can be beautiful, introspective, moving, argumentative. I get all that, I really do. There’s no difference in category here though: beauty is replicable, and so is argument and emotion. I think that software, in the future, could express skepticism in writing when it is given a claim, and learn how to give different weight to conflicting ideas.
But I don’t think computers have to fully do these things. They need to do it just enough for it to make economic sense to employers, to convince those who pay that a cheaper, maybe slightly duller version makes more economic sense than the real thing. That’s true for taxi-drivers, but it’s also true for attorneys, businesspeople, and even doctors (more on that last one later, I hope).
I don’t know the timing of any of this. It could be twenty, or a hundred years off, maybe much more, but computers will be able to do what you do, only more efficiently, without asking for raises and having babies and doing all of those other pesky things employees do.