Economist Tyler Cowen Explains Why AI Won’t Be the Revolution We Expect

AI is a great innovation, but let's not go any further than that just yet

When everything is a revolution, nothing is.

There are two ways we can argue that AI won’t be a revolution, that it won’t change the world as much as we’d like or as much as AI companies, the media, and even most AI bloggers would make us believe. One of those arguments is nonsense; the other is very reasonable.

The first is the anti-AI stance, which I reject despite it’s weirdly common: AI is useless at best and dangerous at worst. Anything that may come out of AI progress, as it's conceived now, is bad for the world—socially, culturally, politically, environmentally, and economically.

I have three reasons to despise this take passionately.

One, it’s so obviously false as to be impossible to be used in a debate except as a weapon, like a sword wielded against an enemy. We could argue that progress has been made in some aspects and not others; that AI is more engineering—even alchemy—than science; that we will eventually backtrack a few years of efforts after finding an unpassable wall; or even that more ethical ways of doing things are possible—and preferable. I empathize with all those arguments and share a couple. That’s very different than saying AI is strictly bad for the world, which is just a strange ideological posture that makes zero sense.

Two, it’s true that AI will bring objectively bad things (job disruption, disinformation, discrimination, cybercrime, etc.). Still, the argument loses all its weight when these harms are conflated with a generalized dismissal of the good things and advances that AI could—and is bringing. To use the same metaphor: all technological innovation is a double-edged sword—fire heats us and our food but can also burn the world—it’s how we use it that matters.

(We shouldn’t forget that innovation emerges from and is steered by specific interests that may not be reflective of universal preferences, but that’s a different and much deeper issue that although worth talking about, in no way denies the value that technology can bring, even if that’s not always the case.)

And three, the one that personally stings; outsiders may mistakenly assume that every argument of the form “AI may not be as revolutionary as some people claim,” defaults to this extreme stance of “AI is worthless.” That’s not true and we’d do well to redefine the narrative to account for the pro-AI, pro-technology, but moderately cautious and historically-grounded version of this idea.

That’s the version I think is true and the one I support—and I’m happy to know that I’m not the only one.