On Taste

In recent years, the rise of artificial intelligence has led to much debate on the impact of AI on various human domains, such as translation, programming, legal research and writing corporate reports. The consensus among many is that while AI excels at handling mechanical tasks, it lacks the human touch known as "taste." Taste can be defined as a sense of aesthetics and judgment derived from experience and intuition, allowing one to discern whether something is good or bad without being able to explain the reasoning behind it.

As a software engineer, the author agrees with the notion that AI-generated code lacks taste. AI now handles code generation, leaving engineers to act as the "taste judges" for individual pieces of code, software architecture, and product and UX decisions. The concept of taste is not new, but discussions on the subject have gained momentum.

In the 1970s, Richard Hamming, a mathematician at Bell Labs, created a course called "The Art of Doing Science and Engineering," with the goal of teaching taste or style. Though he acknowledged that taste cannot be taught through traditional methods, Hamming focused on providing examples instead, primarily drawn from his 30 years in the mathematics department of Bell Telephone Laboratories.

The idea of taste resurfaced when Andrej Karpathy coined the term "vibe coding" in 2025, inspiring memes featuring Rick Rubin. These memes resonated with software engineers who began feeling their coding abilities wane as AI became more prevalent. The author, a software engineer turned entrepreneur, emphasizes the importance of taste in decision-making for products, architecture, organization, and sales.

Taste plays a crucial role in shaping the direction of a company and its offerings, as it involves making judgments about what to build, how to communicate, and how to present information.

The author wrestles with the question of what humans will be left to do as AI becomes more advanced. While AI can generate content faster than humans, the author argues that technique will eventually become obsolete, leaving taste as the primary differentiator. As AI continues to automate tasks, the author believes that some individuals with a knack for taste will have a competitive edge and be able to capitalize on this distinction.

However, the author acknowledges that taste may not be enough to protect everyone for long, and the field may ultimately shift once more.