I was the worst kid in art class. Not the charming, misunderstood-genius kind. The “Ronnie, that’s supposed to be a horse?” kind. The talent never showed up. My first website designs were embarrassing. My current websites are, honestly, still kind of embarrassing.
So when AI tools started spitting out decent layouts and color palettes, I was thrilled. The output wasn’t great. It was fine. Solidly average. And average was a huge upgrade for me.
Here’s the part I keep coming back to though. AI is average. I don’t mean that as an insult. It’s just how the math works.
These models are trained on enormous amounts of text and they learn the statistical patterns of which words tend to follow which other words. When you give one a prompt, it isn’t thinking. It’s picking the most likely next word, over and over, a few thousand times in a row, and you get a paragraph. Rodrigo Sperb put it well: “the most probable, polished, average answer at lightspeed.”
The research is starting to back this up. A 2026 study by Cruces et al. found that AI closed about 75% of the performance gap between workers with less education and workers with more. The less skilled you were going in, the more AI helped. UChicago economist Alex Imas, in his review of the research, said AI “compresses rather than widens the productivity distribution.”
So if you’re below average at something, AI is a gift. It can pull you from the 20th percentile to the 60th or 70th overnight, no practice required. That’s been my experience with design, and it still feels a little magical.
But what about the work you’re already good at? That’s where I think a lot of us are quietly getting worse without noticing.
I’ll use myself as the example. Despite what a few of my high school English teachers might have thought, I actually enjoy writing. I’ve developed a voice, opinions, a rhythm of putting things together that’s mine. It’s not perfect (you’re reading it, you can judge), but it’s specific. It’s me.
If I let AI write this post, it would be fine. Probably smoother than what I’m typing right now. But it would also read like every other AI blog post on the internet, which is to say, like nothing in particular. I don’t want my writing to sound like a statistical average of everyone else’s writing.
I’m writing this post because I feel the pull myself. It’s so easy to let AI take a first pass and then “just clean it up.” You tell yourself you’re still in control, that you’re really the editor. Each time you do it though, you lose a little bit of the muscle. The writing gets a little smoother and a little less yours. I’ve caught myself being happy with average because it showed up fast and looked professional. Not great.
I’d guess the same is true across a bunch of jobs. If you’re in product (hi, that’s me too), AI is great for summarizing user research, drafting first cuts of specs, organizing customer feedback. But please don’t ask it for product strategy. You’ll get a perfectly average strategy that sounds exactly like the strategy at every other company in your space. The actual reason you got hired is to see what other people don’t see. Statistical models can’t do that. By definition, they give you the most common answer.
Same general idea if you’re a marketer, or a lawyer, or a developer. Lean on AI for the grindy stuff where “good enough” really is good enough. Document review. Repurposing content. Generating predictable, well-documented code. But the parts where your judgment has consequences, where the right call isn’t the most common pattern? That’s where average is actually dangerous.
The thing I worry about isn’t that AI will replace any of us. It’s that our work quietly drops from above average to average, and nobody notices, including ourselves. It just gets a little more forgettable. A little more like everyone else’s.
So I’m trying to be more deliberate about it. Know what I’m actually good at, and do that work myself. Use AI for the rest.
What are you better than average at? And are you still doing it yourself?
Leave a Reply