Looking back on that initial site build, I used Cursor and Claude to build my personal website from scratch. At the time, it used Hugo, multilingual support, automatic translation, GitHub Actions CI/CD, and real-time Fitbit data integration. It went from concept to launch in under two weeks.
Based on how I worked at the time, I estimate that three years earlier the same undertaking would have taken me roughly three months, and I would have had to outsource part of it.
But one thing didn’t get faster: deciding what this website “should look like.”
Not the question of visual design. A more fundamental one: who is this website for? What should it convey? Which features are core and which are distractions? How many thematic pillars should the articles be divided into? What is the relationship between each pillar? After a reader finishes an article, where should they be guided next?
AI can help me organize reader needs, compare options, and even point out conflicts between my goals. But I still need to state clearly who I am and what I believe matters. I also remain responsible for the direction I choose. That kind of judgment is what I mean here by taste.
Taste Isn’t What You Think It Is
The word “taste” carries a sense of elitism, as if it’s saying “I understand wine, you don’t.” But the taste I want to discuss has nothing to do with elitism.
The design I care about does more than make things look good. It should make them fit their use: whom they serve, what problem they solve, and what constraints people will encounter.
“Making things right.” That’s the essence of taste.
More concretely, taste is the intersection of three abilities: discernment (seeing which option among many is the right one), the power of negation (the courage to say “we won’t do this”), and a sense of context (understanding why, in this particular situation, this choice is the right one).
Discernment can be accumulated through experience. The power of negation requires courage and judgment. But the sense of context is the hardest. It requires you to simultaneously understand technical constraints, user needs, business logic, and cultural background, and then find the optimal solution at the intersection of these dimensions.
When the design principles, use case, and resource constraints are made explicit, AI can help score and rank options, and it can suggest reasons not to pursue something. Whether that ranking is useful still depends on whether the conditions are complete, the assumptions are reasonable, and the result works in practice. My job is to supply context, check the recommendation, and take responsibility for the tradeoff.
How My Taste Was Trained
Honestly, I’m not someone with innate taste. My taste was forged by falling into pits.
In my years as a consultant, I saw too many products that were “technically perfect but commercially meaningless.” A team would spend half a year polishing a feature, only to find after launch that users didn’t care at all. The problem wasn’t technical execution; it was that the right question was never asked at the start: “Whose pain point, exactly, does this feature solve?”
Later, working in the circular economy, I learned another layer of taste: a sense of fit. The same metal-recycling technology could be a revolutionary improvement on Factory A’s production line and a superfluous complication at Factory B. The difference wasn’t in the technology itself, but in whether you could judge “whether this solution suits this scenario.”
This connects unexpectedly with my experience of theological training. Fifteen years of theological background taught me one thing: the same passage of scripture can carry entirely different meanings in different contexts, and the quality of interpretation depends on the depth of your understanding of context. This “context sensitivity” later became the underlying ability behind all my judgments, whether choosing a technical architecture, setting content strategy, or deciding the angle of an article.
Now, building systems with AI, I find the value of taste has become more evident, not less important.
Here is a concrete example. When I built the debate engine, there were many design choices: how many models to use, how many rounds of dialogue to run, and where to place fact-checking. Each combination still had to be checked for cost, limitations, and results. At the time, I chose four models, dual debate-and-collaboration modes, and Perplexity for the final fact-check based on dozens of experiments I had run, the output tendencies I had observed across models, and the quality I wanted. That was the configuration at the time; model performance changes, and a final fact-check does not guarantee that the output is correct.
AI wrote every line of code in the engine. But what the engine “should look like” was decided by taste.
What Vibe Coding Really Means
Many people understand Vibe Coding as “you don’t need to seriously write code anymore, just chat with AI.” This understanding is wildly off.
The lesson I take from Vibe Coding is this: making some prototypes faster does not make tradeoffs unimportant. Deployment, testing, maintenance, security, and data quality still carry costs, and decision quality is not the only source of differences in outcomes.
In After Code Gets Cheaper: From Vibe Coding to AI Agents, Who Is Responsible for Delivery?, I separate the cost of producing a first draft from the cost of confirming that it works and maintaining it after delivery. That piece is about the boundary between experimentation and delivery; this one asks how an individual decides what is worth building.
One important factor is the quality of the instructions, which in turn depends on how well the problem is understood. Reliable data, suitable tools, engineering checks, and user feedback also need to support the work.
When I describe a requirement in natural language, I try to state both the intent and the technical conditions clearly. “Receive scheduling data in JSON, validate the fields, and store it in the database” describes the technical work, but the resulting code still needs testing. “Let one person manage publishing across eight social platforms, with reliability taking priority over flexibility” describes the direction of the tradeoff. The two complement each other; a clear goal does not guarantee an error-free implementation.
The first leans toward engineering specifications, while the second leans toward what I call taste. Both matter, and they often overlap.
Why the Humanities Suddenly Became Important
In my own experience collaborating with AI, I have come to place more value on the ability to articulate an idea clearly. This is not a ranking of technical skill against communication; both affect the collaboration.
This sounds simple, but “articulating clearly” is actually an extremely complex ability. It requires you to first think clearly about what you want (self-awareness), then express it in a way the other party can understand (communication ability), while anticipating where they might misunderstand and clarifying in advance (empathy), and finally identifying within their response which parts are right and which need correction (critical thinking).
These four abilities, self-awareness, communication, empathy, critical thinking, are all at the core of humanities training. Rhetoric teaches you how to express precisely. Philosophy teaches you how to deconstruct problems. Literature teaches you how to understand context. History teaches you how to distill judgment from cases.
In my own experience, theological training has helped my collaboration with AI far more than any programming language. Because the core training of theology is precisely this: facing a complex text, finding the most reasonable, most responsible interpretation among multiple possibilities. This is, in essence, the same thing you need to do when facing AI’s output.
This is also why I believe the “post-code era” is not the end of engineers but a renaissance of the humanities.
Can Taste Be Cultivated?
Yes. But not by “taking a course in taste.”
Taste comes from three sources: abundant input (seeing enough good and bad things), cross-domain connections (finding a common framework of judgment across experiences in different fields), and repeated practice and feedback (making choices, bearing the consequences, correcting your judgment).
Vitsœ’s Dieter Rams: ten principles for good design includes usefulness, understandability, honesty, and attention to detail in design judgment. To me, it is a reminder that taste can be supported with reasons rather than reduced to “I like it.”
For me, cultivating taste has a very concrete method: deliberately practicing “why not to do it.” Every time you make a decision, record not only what you chose but also what you gave up and why. Over time, you’ll find your framework of judgment becoming increasingly clear.
That’s exactly what I did running my personal website. Before writing each article, I’d first list three to five possible angles, then eliminate them one by one until only the most valuable one remained. The eliminated angles weren’t bad; they just weren’t the most “right” choice at this point in time, for this group of readers, within the context of existing articles.
This process is the muscle memory of taste.
Keeping Judgment and Responsibility in the Collaboration
In AI Agents vs. Agentic AI, I discuss how human–AI collaboration still needs architectural choices. In Neuralese and AI Oversight, I also consider how model processes that are difficult to read can make oversight harder. But being able to read a process does not make its output correct, and being unable to read it does not make every form of external verification useless.
The intersection of these two issues is taste.
Taste helps determine what I ask AI to do and which acceptance criteria I use. But supervision cannot rest on personal preference alone; it also needs testing, data checks, and user feedback.
AI can also help discover context, propose goals, and challenge assumptions. I still need to confirm whom the work serves, which resources are limited, and which values are worth preserving. Accepting a recommendation does not mean handing over responsibility for the decision.
Code can be copied, and models can be trained. What remains worth practicing, for me, is choosing what to build, deciding what to leave out, and being able to explain both the reasons and the results.
Clarifying the Conditions and Choices
AI speeds up making things; that makes our tradeoffs even more worth explaining.
AI can compare options, but the direction we choose still needs a reason.
Intent and technical specifications complement each other; they do not guarantee error-free implementation.
Taste takes experience, testing, and user feedback.
Record what you choose not to do, then use the results to revise your judgment.
State the goal clearly, and take responsibility for the option you choose.
Revised November 5, 2026. The site build and model configuration describe my work at that time; three months is my own estimate. This essay uses taste to discuss tradeoffs and responsibility, and its design principles draw on the Vitsœ page linked above.


💬 Comments
Loading...