When “Good Enough” Becomes the Gold Standard

AI is moving fast. Everything around it feels temporary: tools, workflows, even the way we talk about what matters. In all this motion, one observation keeps getting sharper: the risk of mediocrity.

The risk emerges from the interaction between human behavior and AI capabilities. AI is exceptionally good at lowering the bar on cognitive work. Humans are exceptionally good at optimizing for utility. Put the two together, and a familiar pattern begins to form.

When a tool delivers 80% of the result for 20% of the effort, stopping there is not a failure of ambition or discipline. It is a rational decision. The remaining 20% usually demands disproportionate time, attention, and care, while producing marginal gains that are hard to justify under real-world constraints, deadlines, competition, and incentive structures.

Fast forward a few years — what happens when AI starts delivering something closer to 90% at 10% effort? Further improvements become even less of an obvious choice. The additional investment begins to look inefficient, the upside unclear, and in many cases misaligned with organizational or shareholder interests. Excellence, at that point, stops feeling like a default expectation and becomes a niche preference.

None of this is new; every wave of modernization has forced the same trade. Industrialization brought abundance and accessibility, often at the expense of quality. Ask someone in their 50s about the taste of food in diners thirty years ago, or about the quality of handmade leather goods from a time before efficiency and scale became the priority.

In tech companies, this dynamic is already at work. AI is “blurring the borders between roles”. Designers, product managers, and developers all prototype, talk to customers, and commit code. In some cases, a single talented person juggles all of this without dropping any balls. But what happens on average? Would anyone mind if the quality slightly drops? Not when doing more with less, and velocity is the new goal. For the sake of the argument, let’s assume the average quality doesn’t drop; it even slightly lifted — still, the point is, excellence is a luxury that no one wants to afford.

What changes with AI is not the tradeoff itself, but the territory. Previous revolutions automated hands, muscles, and manufacturing. This one automates the early stages of thinking. Drafting, researching, exploring, and structuring; activities that once imposed natural friction, now arrive almost for free.

That shift is already visible in everyday behavior. Conversations with AI that feel profound in the moment, then quietly abandoned. Research tasks delegated, results returned, and never fully read. Tokenized ideas that feel complete as soon as they are articulated, without ever being tested, debated, or challenged.

The danger is not laziness. It is comfort. When everything comes easily, it becomes easier to settle, for the summary instead of the substance, for momentum instead of depth, for good enough instead of done properly. Over time, this recalibrates the baseline. Half-baked feels complete. Unfinished feels sufficient. Depth becomes the fine print that no one wants.

Under these conditions, excellence is not eliminated, but it is no longer structurally encouraged. It requires resisting efficiency, overinvesting where the returns are unclear, and continuing well past the point where stopping would be perfectly defensible.

There is a scene in Whiplash, a 2014 film, where Terence Fletcher, played by J. K. Simmons, explains why he believes the most damaging words in the English language are “good job.” Not because failure should be punished, but because premature approval puts motivation to sleep. “Good enough,” in his view, is how potential quietly caps itself. It is an extreme philosophy, deliberately uncomfortable, but it captures something real about how easily the drive toward excellence can be neutralized, not by criticism, but by satisfaction.

This is not a dark prophecy, but rather an attempt to shed light on a manageable risk. AI can do an enormous amount of good, but it requires a certain kind of attention, intention, and a willingness to resist some very natural inclinations. The effects will not be uniform. Some people will drift more than others, and the outcome will naturally distribute itself along a curve. The point of raising this is not to predict where the curve will land, but to influence where one chooses to stand on it. And if enough people are deliberate about that choice, the average itself can still move in the right direction, and save excellence, the current standard, from becoming a niche.

(Fletcher said it better — here)