In the Age of AI, Is Taste Becoming More Important Than the Ability to Create?

Artificial intelligence can produce a draft of a text, a musical composition or several image variations in seconds. That does not mean the creative process has been fully automated. On the contrary, as the number of possible results grows, a more important question emerges: which one is worth finishing, and why?

So ‘taste or the ability to create?’ may be a false choice. AI makes certain technical actions easier, but creative work is more than execution. It includes forming an idea, understanding the material, choosing a direction, assessing the result and taking responsibility for the finished work.

More work, more choices

A study published in PNAS Nexus analysed more than four million works and the activity of over 50,000 users on a text-to-image platform. The researchers recorded an increase in the number of works produced and in likes per view after people began using AI. At the same time, average content novelty declined, as did visual novelty among both average and top results. The authors conclude that finding ideas and filtering results become important skills in this process.

The study does not measure ‘taste’ directly. It shows that AI can increase the volume of work, while finding a distinctive and unexpected solution remains a challenge. A large number of variations does not, by itself, make a work interesting or faithful to the creator’s original intention.

AI can help individuals while making outputs more alike

An experiment involving short stories included 293 writers. One group could use one or more opening ideas generated with GPT-4; the other wrote without AI. Evaluators judged stories by participants with access to AI ideas as more novel and useful. The advantage was especially pronounced among participants who had scored lower on an earlier creativity task. However, the AI-assisted stories were more similar to one another and often drew on ideas suggested by the model. The experiment concerned short fiction, not every creative field.

A 2025 meta-analysis bringing together 28 studies and 8,214 participants also found a two-sided result: people working with AI performed better on average in creative tasks than people working without it, but the diversity of their ideas decreased.

These findings do not mean that every AI-assisted work is alike or that an individual creator becomes less creative. They raise a more specific question: when many people use similar systems, how can we preserve distinct perspectives and local experience?

What do we mean by ‘taste’?

Here, taste is not simply a matter of ‘I like it’ or ‘I don’t’. In a creative process, it can mean the ability to identify what a work is trying to say; recognise which result has meaning; assess it within a particular genre and context; and make a choice whose reasoning you can explain.

But taste is not a neutral measure shared equally by everyone. It is shaped by education, experience, culture and the environment in which a person lives. Declaring one standard of ‘good taste’ can therefore act as a cultural gatekeeping mechanism: who decides which styles have value, whose experience is visible and which cultural influences are preserved?

Craft still matters

AI speeding up some stages of production does not mean creators no longer need knowledge of their field. A musician needs to understand musical form, a designer composition and visual hierarchy, and a writer language, character and rhythm. Without that knowledge, it is difficult to tell when an AI suggestion works and when it merely looks convincing.

A meta-analysis of 106 experiments found that people working with AI achieved better results on average than people working alone. But the combined result did not, on average, outperform whichever side — the human or the AI — performed better independently.

Results also differed by task: collaboration looked more promising for content creation than for decision-making. This suggests that AI’s usefulness depends not only on the tool, but also on what a person knows and the role they play in the process.

Legal systems also distinguish between material generated by AI and a person’s creative contribution. The U.S. Copyright Office says that work made with AI may qualify for copyright protection when it contains sufficient human authorship — for example, in the selection or arrangement of material, or in later modifications.

In one specific case, the Beijing Internet Court recognised copyright in an AI-generated image. The court took into account that the creator selected and revised prompts, adjusted parameters and chose the final version.

The answer is not simply ‘taste’

In the age of AI, creative value is unlikely to shift neatly from one skill to another. Stages of the process that often went unseen behind production are becoming more visible: choosing an idea, setting a direction, reviewing the result, rejecting unnecessary versions and taking responsibility for the final decision.

The question may be put this way: when creation becomes easier, what helps a work avoid getting lost among similar results? Taste may become an important advantage, but it cannot develop without experience, professional knowledge and continued practice.