As AI expands what teams can produce, creative leaders face a new question: who decides what is worth making? Taste can work as an operating advantage when it means tested, domain-specific judgment. Calling it a lasting moat is a strategic metaphor, not a proven fact. AI and creativity meet at this point of selection.
People often treat taste as a vague gift. In practice, it is a skill you can define, practice, and audit. It means setting criteria, seeing trade-offs, using domain knowledge, explaining a choice, testing it against its job and audience, and owning the outcome. This view turns a fuzzy quality into a repeatable process any creative leader can build.
Why creative judgment matters as AI expands output
When anyone can make many options in minutes, the bottleneck often shifts from making to choosing. In many settings, quality work still takes real skill and cycles. But the balance is shifting in some creative fields.
Research shows the shift is real but subtle. Access to AI ideas improved average evaluations on a short-story task, especially for less-creative writers. At the same time, the stories became more similar to one another (Doshi & Hauser, 2024). This is a task-specific result. It points to a risk that easy idea access can narrow range in that kind of task.
GPT-4 responses scored above a sample of 151 people on originality and detail across three divergent-thinking tasks (Hubert et al., 2024). That finding has a clear limit. Divergent-task scores measure creative potential under lab conditions. They do not measure full creative achievement, taste, market fit, or accountable choices. Passing a lab test is not the same as shipping work that connects with a real audience.
What taste means in practice
Taste, defined in clear terms, is the ability to make defensible creative choices. It has six parts: setting criteria before seeing options, seeing trade-offs, using domain knowledge, explaining a choice in plain language, testing it against its job and audience, and owning the outcome. This separates taste from mere preference. A preference is a feeling. Taste is a judgment you can defend and test.
Selection involves real trade-offs. Across idea-selection studies, people tended to favor feasible ideas at the expense of originality. Explicit originality instructions improved originality selection but reduced satisfaction and rated effectiveness (Rietzschel et al., 2010). Creative judgment is not a flawless instinct. It is a balancing act among competing values.
Can intuition be trusted for creative choices
Intuition has a role, but its trustworthiness depends on conditions. Intuitive skill is most likely when the setting has learnable patterns and the person gets enough practice and feedback. Confidence alone is not proof that an intuitive judgment is accurate (Kahneman & Klein, 2009). This is a sobering finding. A seasoned creative director's gut feeling may hold value, but only with steady patterns and clear feedback over time.
Skill can help spot novel ideas, but the decision context matters. Expertise supported recognition and selection of novel ideas in one study. Taking a formal decision-maker role weakened that observed effect (Beretta et al., 2025).
How to build a practical judgment protocol
A judgment protocol turns taste into a repeatable routine. It organizes your thinking so you can see where your judgment is strong and where it may be biased.
Apply a clear protocol each time:
- Define the job and audience before seeing any options.
- Set weighted criteria in advance.
- Gather strong and weak reference pairs to calibrate.
- Generate truly different directions.
- Separate making from first review when useful.
- Score and explain trade-offs.
- Test with fitting users or proof.
- Keep a decision log and review results later.
Setting criteria before seeing options may stop a strong first idea from anchoring you. Reference pairs give concrete comparison points. Separating making from review may stop you from falling in love with your own output. These are practices worth testing, not guaranteed fixes.
The scorecard is a compact tool for this protocol.
| Criterion | What it asks |
|---|---|
| Fit | Does it answer the brief and suit the audience? |
| Clarity | Can the audience grasp the idea quickly? |
| Distinctiveness | Does it stand apart from obvious options? |
| Usefulness | Does it serve the strategic job? |
| Craft | Is the execution well-built and polished? |
| Evidence or rights risk | Are claims checked and assets cleared? |
Weights depend on the task. A technical whitepaper needs accuracy and clarity most of all. A brand campaign needs distinctness and emotional fit. One universal rubric does not fit all projects.
Consider a neutral case. A team has three campaign directions for a financial-services brand. One stresses stability and trust. One focuses on digital change. One is a bold, funny take on daily money stress. Using a weighted scorecard, the stability angle scores high on fit and craft but low on distinctness. The funny angle scores high on distinctness but carries more tone risk. The change angle sits in the middle. No option is wrong. The choice depends on the audience and the strategic job. The scorecard makes the trade-off visible and discussable.
For leaders, a protocol clarifies when to involve others. Single taste moves fast but carries personal bias. A team may be slower but can correct blind spots, if the process allows dissent. The risk is groupthink dressed as agreement. A protocol that needs separate scoring before discussion may surface real dissent.
AI can help screen in a structured way. Large language models reached moderate alignment with expert idea evaluations only when given a base framework and spectrum-based few-shot examples (Kranzle & Sharratt, 2025). This supports structured AI-assisted screening, not a claim that models possess universal taste. Use AI as a first-pass filter, not a final judge. It can sort options against your stated criteria and flag clear risks.
For early-career creatives, keep building skill while using AI as a lever. Practice the parts of craft that still need human feedback loops. Study strong and weak work with focus. Ask for specific feedback on your choices. The judgment habit forms through doing, not through passive use of made output.
What must stay accountable even as tools improve
Even as review tools grow sharper, ownership does not shift. An organization cannot delegate responsibility for a final decision to a tool. Someone must own the result, explain the thinking, and answer for the outcome. This holds whether the choice was made by a person, an AI, or a mixed process.
Ownership has three parts. First, traceability: you should be able to rebuild why a choice was made. Second, clarity: you should explain it to others in plain language. Third, learning: you should review results and update your criteria for next time.
A decision log backs all three. Record the options, weights, scores, and thinking. Return weeks later to see what held up. This turns judgment into a cycle of result and change, not a one-time act.
Key takeaways
- Taste is a clear skill: set criteria, weigh trade-offs, explain choices, test them, and own the result.
- AI can lift average output in specific tasks but may narrow range. Guard variety with care.
- Intuition is more likely to be reliable in learnable settings with practice and feedback.
- Test a judgment protocol with pre-set criteria, reference pairs, and separate scoring.
- Use AI for structured screening, not final accountable choices.
- Keep a decision log to check and improve your creative judgment over time.
Frequently asked questions
References
- Beretta, M., Deichmann, D., Frederiksen, L., & Stam, D. (2025). Do you see what I see? How expertise and a decision-maker role influence the recognition and selection of novel ideas. Research Policy, 54(1), 105139.
- Doshi, A. R., & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28), eadn5290.
- Hubert, K. F., Awa, K. N., & Zabelina, D. L. (2024). The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks. Scientific Reports, 14, 3440.
- Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: A failure to disagree. American Psychologist, 64(6), 515-526.
- Kranzle, T., & Sharratt, K. (2025). Evaluating creative output with generative artificial intelligence: Comparing GPT models and human experts in idea evaluation. Creativity and Innovation Management, 34(4), 991-1012.
- Rietzschel, E. F., Nijstad, B. A., & Stroebe, W. (2010). The selection of creative ideas after individual idea generation: Choosing between creativity and impact. British Journal of Psychology, 101(1), 47-68.
This article is for informational and educational purposes only and does not constitute financial, legal, tax, medical, or professional advice. Individual results vary.