When a prominent tech CEO was asked what would remain irreplaceable as AI transforms software development, his answer surprised many: “taste.” Not technical skill, not coding ability, but the seemingly subjective capacity to discern what’s worth building and how it should work.
This reveals a profound paradox reshaping all knowledge work. As our tools become infinitely capable, our most human qualities become exponentially more valuable. Yet most of us spend our time honing the very skills that machines are rapidly mastering.
The concept of taste isn’t new. Ancient Greek philosophers had a word for it: ‘phronesis’ (φρόνησις) – practical wisdom that allows you to make good judgments in uncertain situations. They distinguished it from both technical knowledge and theoretical understanding. It was the ability to know not just how to do something or what was true, but what was worth doing and when to do it.
For centuries, this capacity was bundled with technical execution. A master craftsman needed both the skill to carve wood and the judgment to know what was beautiful. A writer needed both grammar mastery and narrative instinct. A strategist needed both analytical capability and situational awareness.
AI is now ‘unbundling’ these capacities. Programming has historically required what we might call “human compilation,” taking a clear vision and laboriously translating it into code. The programmer had to be both visionary and translator, both architect and construction worker. As AI handles more of the translation layer, what remains is pure intention: knowing what you want to build, how it should feel, and why it matters.
This same unbundling is happening across knowledge work. Marketers focus less on crafting copy, more on understanding human psychology. Lawyers spend less time on document review, more on strategic positioning. Analysts do less data processing, more on pattern recognition and narrative creation. Teachers handle fewer administrative tasks, focusing more on inspiration and connection.
Taste operates on multiple levels. There’s immediate aesthetic judgment, the gut feeling about whether something looks or feels right. There’s functional taste, understanding how things should work and what creates the right user experience. There’s strategic taste, knowing what problems are worth solving and whether something should exist at all. There’s cultural taste, sensing what resonates with people and will matter to them.
Most people develop surface-level taste through exposure and practice. But the deeper levels, the ones becoming most valuable, require different kinds of cultivation. This creates a paradox: How do you develop taste in a world where AI can execute most ideas instantly?
The traditional path was apprenticeship, developing judgment through countless hours of execution. You learned what worked by building things that didn’t work. You developed aesthetic sense by making ugly things and slowly improving. But if AI can now execute your ideas immediately, you lose this feedback loop. You can generate impressive demonstrations without understanding why they’re impressive or whether they should exist at all.
This creates what researchers call “the judgment gap.” People with underdeveloped taste suddenly have access to sophisticated execution, while people with refined judgment may not know how to leverage new tools effectively.
The solution isn’t to avoid AI tools, but to use them deliberately for taste development. Use AI to quickly test many variations, focusing your attention on discerning which ones work and why. The speed allows you to see patterns in quality that would take months to recognize manually.
Give AI specific constraints that force you to make judgment calls. Instead of “make this better,” try “make this appeal to someone skeptical of technology” or “optimize for clarity over cleverness.” Use AI to help you explore how solutions work in adjacent fields, developing your ability to recognize deeper principles that transcend specific implementations.
Regularly create terrible versions of things intentionally, then improve them. The contrast sharpens your ability to distinguish good from great. The goal isn’t to compete with AI on execution speed, but to develop the judgment that guides what gets executed.
We’re entering a decade where our ability to build will be magnified beyond anything in human history. But this magnification is multiplicative, not additive. Poor taste multiplied by infinite capability creates infinite mediocrity. Refined judgment multiplied by AI execution creates exponential value.
This means the stakes for developing taste have never been higher. In a world where anyone can create anything, the ability to discern what should be created becomes the ultimate competitive advantage. The question isn’t whether you can execute your ideas anymore. The question is whether your ideas are worth executing. That’s a judgment only you can make, but only if you’ve cultivated the taste to make it well.
In your current work, what percentage of your time is spent on execution versus deciding what’s worth executing? How has this ratio changed in the past year as AI tools have become more capable? Think of someone in your field whose judgment you deeply respect, someone who consistently chooses the right problems to solve or creates things that resonate. What specific patterns can you identify in their decision-making that you could begin practicing? If AI could handle 80% of your current technical tasks within two years, what forms of judgment or discernment would you need to develop to become more valuable, not less? What’s one small experiment you could start this week to begin cultivating that capacity?
