When people think of craftsmanship, they don't think of AI. Maybe they should.
I heard something this week that stopped me. Ben Affleck, of all people, put it better than most people writing about AI have.
Craftsman is knowing how to work. Art is knowing when to stop.
He said it while talking about AI in filmmaking. His point was that AI is a craftsman at best, it's good at imitating technique, but it doesn't know when something is actually done. I haven't stopped thinking about it since, mostly because it describes exactly what I keep catching myself doing with AI tools all week.
The urge to keep going
You know that thing where you've already got a good draft, but you ask for "just one more pass" anyway? That's the trap. It feels productive. It feels like you're being thorough. But a lot of the time you're just sanding a surface that was already smooth.
Knowing when something is finished is its own skill, and honestly it might be the harder one. It takes actually trusting your own judgment enough to say "this is good, I'm done," instead of letting the tool talk you into round six.
There's a name for this, and it's not really about AI
Turns out psychology already has a term for exactly this pattern. Researchers split people into two types: satisficers, who pick the first option that meets their bar and move on, and maximizers, who keep comparing and optimizing even after they've found something good, because they're chasing the best possible option, not just a good one.
The research shows maximizers report less satisfaction and more regret than satisficers, even when their outcomes are just as good, sometimes better. All that extra searching doesn't pay off emotionally. It just stretches out the discomfort.
AI quietly turns all of us into maximizers. In real life, "let me try one more version" has friction, it takes time, effort, energy, so most people naturally stop once something's good enough. AI removes that friction completely. One more draft costs a few seconds, so the maximizing instinct runs wild, even though the research says that chase leaves people feeling worse, not better.
AI will never tell you to stop
A person gets tired. At some point you run out of steam, or the page just stops giving you anything new to fix. AI doesn't have that limit. Ask for a fourth draft and it'll happily give you one, even if the second draft was already right.
So the stopping point has to come from you. The tool has no opinion on when enough is enough, it'll keep going as long as you let it.
I caught myself doing this earlier this week, asking for another round on something that was already good, not because it needed it, but because asking felt easier than deciding I was done.
Maybe the real skill is knowing when to stop
Most conversations about AI right now are about speed. How much you can crank out, how fast. Way less of it is about restraint, about knowing when to actually put the thing down. Which is funny, because restraint has always been the harder half of any craft. Anyone can add one more layer. Knowing which layer is the last one, that's the actual skill.
Where that leaves me
Fewer "one more passes." More just asking myself honestly, is this actually done. Less letting the tool's endless patience set my pace, more trusting my own gut on when something's finished.
The craft was never just in the doing. It's in knowing where to stop.
Happy Learning,
KP
Keep reading
AI & FluencyAug 12, 2026 · 6:58 AM2 min read
The reason AI "forgets" you has nothing to do with AI
It's not a glitch. It's the same reason your memory works differently underwater than on land.
AI & FluencyJul 31, 2026 · 9:52 PM3 min read
AI drafts. It doesn't decide.
The best tools speed up production. They still can't tell you if it's any good.
AI & FluencyJun 17, 2026 · 11:23 AM3 min read
What a good AI Fluency curriculum actually looks like
Fluency isn't a tool skill. It's a judgment skill. Most training gets that backwards.