Taste in the age of AI and LLMs

A debate is splitting Silicon Valley: Is "taste" — the ability to judge good from bad, interesting from boring — the most important skill now that AI can generate almost anything on command?

One camp, led by voices like Sari Azout, argues that AI's creative power makes taste more valuable, not less (Source 3). The logic: ChatGPT and Claude can write a product description, design a logo, or draft code in seconds. But they have no opinion on whether that output is actually good. They're like a chef who can cook any dish perfectly but has no idea what should be on the menu. As context windows expanded from ~1,500 words two years ago to 1.5 million words today, you can now feed AI an entire reference library — your favorite essays, quotes, design examples — and it will generate work in that style (Source 3). Your curated collection becomes the AI's creative DNA.

The counterargument is sharper: taste only matters if you're the one making final decisions, which describes maybe 5% of workers (Source 2). Most people aren't designing the menu — they're running the kitchen. And kitchens don't run on taste; they run on systems. The "taste is everything" advice, critics say, is harmful because it tells the majority of workers to develop a skill they'll rarely use professionally. What matters more is systems thinking: understanding how AI fits into workflows, how to route tasks between human and machine, how to quality-check outputs at scale.

Behind this debate is a deeper anxiety: tech workers are watching AI automate their skills and scrambling to identify something machines can't do (Sources 5, 6). "Taste" sounds appealing because it feels human, subjective, uncodifiable. But as one skeptic points out, if taste is just pattern recognition over many examples, AI is already better at it than most humans (Source 4). The question isn't whether AI has taste — it's whether human taste remains valuable when AI can analyze 10,000 design trends in milliseconds.

The real story here is that "taste" has become a class marker in tech. If you're a founder, designer, or executive — someone who decides what gets built and shipped — taste absolutely matters. You're the one saying "this landing page feels off" or "that brand voice is wrong." But if you're an engineer implementing features, a marketer running campaigns someone else designed, or an analyst building dashboards to spec, taste is a luxury skill. You need it occasionally, but workflow design and quality systems matter more day-to-day.

Here's the uncomfortable truth: the "taste matters most" advice is spreading because it's flattering to the people saying it. Creative directors and product leaders want to believe their judgment is irreplaceable. And it might be — for them. But pushing this as universal career advice creates a new anxiety: workers now think they need to become connoisseurs of everything to stay employable, when what most jobs actually require is knowing how to get consistent output from AI tools and catch the 10% of cases where they hallucinate or miss the brief.

The practical middle ground: Taste matters in proportion to your decision-making authority. If you choose what to build, invest in taste — keep a swipe file, study great work, develop opinions. If you execute what others decide, invest in systems thinking — learn to spot patterns in AI failures, design quality checks, automate repetitive judgment calls. Both skills have value. But only one is rare enough to be a competitive advantage, and it's not the one getting all the think pieces.