The Forgotten Knuckleball: A parable for the future of creativity in the age of AI

Nobody uses the knuckleball anymore.

I don't follow baseball, but according to at least some people who do, the knuckleball is one of the most powerful pitches in the game.

This is because, most of the time, balls have spin, which gives them a predictable curve - fastballs go straight because of backspin, curveballs curve because of side-spin.

Knuckleballs don't spin, so they just do whatever they like; often changing their mind several times in the split second it takes to go from pitcher's hand to the strike zone.

Even the best batter facing a good knuckleball will look like an amateur, as they swing at where the ball was supposed to be before it decided to wander off somewhere totally different.

But if the knuckleball is so good, why would you pitch anything else?

The problem is that knuckleballs aren't just the hardest ball to hit, they are also the hardest ball to pitch. If you attempt to pitch a knuckleball and fail, you just handed the batter a home-run.

Of the thousands of pitchers professional baseball has seen over the decades, fewer than a hundred have had the skills to reliably pitch them every time they tried. So coaches and managers have, over the decades, decided that it's better to focus training time on honing easier skills which don't need top talent to deliver good performance.

And this is what leads us to today, where not a single pitcher can be trusted to throw the most effective pitch in the game. Even if they wanted to learn, they couldn't because those guys with that skill are either retired or dead.

Sure, we know the theory of the thing, but there's a long way between theory and practice. Maybe if someone spent their entire life trying to rediscover the art, they could, but not if they wanted to focus on anything else at the same time – like being a professional athlete. The more forgotten a thing, the greater the investment required to retrieve it.

If this makes me sound like I want the knuckleball to return – I don't. I don't care about the knuckleball or anything baseball-related. But I care about what it demonstrates, and that is this: under the right circumstances, we can rationally and deliberately forget a skill, even one which was capable of delivering great results.

Which brings us to creativity.

Generative AI has shifted something.

It has made it possible to reliably generate "good enough" creative content.

There is less reason now to teach someone to be creative. There is less reason for someone to teach themselves to be creative. There is less reason to pay someone to be creative.

I'm certainly not saying there's no value in trying to foster, become, or hire good human creative professionals. There is immense value in it. It is good for the individual and their mind, and it is good for society and culture. It is also good commercially to have artful creative content, powerful, original storytelling, and clear concise prose. All of that has value. But it also comes at a cost.

Training, finding, briefing, shaping, and managing creatives – some of whom will be very good, others of whom will only be acceptable – costs significantly more than zero dollars.

The cost of doing all that for generative AI is not significantly greater than zero dollars. The only difference at the moment is that AI isn't very good, and no one is really set up to use it properly yet. This is not surprising because it has only touched the mainstream in the last year or so. Perhaps this is comparable to 1996 in the development of the internet, as a few nerds were wondering whether this would be good or bad for Yellow Pages and the Post Office.

Most people agree that a lot of creative work is going to be replaced by generative AI.

The consensus at the moment is: "AI will do the churn while humans will do the good bits. The truly talented have nothing to fear. And if you're a human who wasn't good enough to do anything except churn, that's no great loss". But that mistakenly assumes that it will be worth preserving the best of something once the base of it is gone. Knuckleballs didn't stop being good. They just weren't worth the extra effort.

My pessimistic prediction is that in a world where it is easy to produce enormous amounts of rubbish – creative content will stop being valuable to both the people who pay for it to be produced and the people who consume it. Why would anyone want to read anything when nearly everything is the legible slurry pushed out by ChatGPT? And then, why would you pay for a writer when no one really wants to read anything anyway?

Same for visual, audio, film, and illustration. In every case, a feedback loop is created.

It costs less to produce crap, so more crap gets produced, so the audience gets turned off by all the crap, so you have to make more cheap crap more cheaply, and so on. It's a rising tide that never retreats.

The optimistic counter is that demand for the sort of creativity which nearly all of us consider truly valuable in all its aspects – the insightful, meaningful stories, the revolutionary innovations, the growing human understanding, and the fact of its existence as a skill at all – might not be correlated to the fate of that creativity which generative AI will swallow.

Even better, it might also be the case that they are negatively correlated. We may enjoy a future where audiences become connoisseurs of human vs AI creativity, just as the proliferation of fast food prompted a proliferation of foodie culture into mass popular culture.

I know nothing more about the future than baseball – that is to say, not a lot and only what other people have told me. You will have your own feelings and predictions.

I want creativity to flourish and thrive. I want there to be lots of creatives getting paid good sums. But when I read the opinions of the confident, I'm not reassured. They sound like people who think that just because something's the best, then that means it's good enough to survive.