HUMAN CREATIVITY IS NOT OPTIONAL. IT IS THE PRODUCT.

Ninety-six percent. That is the number of architects surveyed in the 2025 RIBA AI Report who rejected the idea that human creativity is no longer needed for building design. Not a fringe position. Not a sentimental one. A near-unanimous professional verdict, issued in the same year that AI adoption inside those same practices jumped from 41% to 59%. Both facts are true at once. That is the only honest starting point. The question being debated in studios right now is not whether AI can generate a floor plan, a render, or a materials palette. It can. The question is whether that output means anything without a human deciding what it should feel like to stand inside the building it describes. Those are different questions. Almost everyone conflates them. ## 59% Adoption, 4% Belief That Humans Are Finished The RIBA AI Report 2025 is the most precise document available on this subject, and its internal tension is instructive. Architecture practices using AI rose from 41% in 2024 to 59% in 2025, with large practices exceeding 80% adoption. The tools are embedded: early design visualisation, specification writing, environmental modelling. The efficiency gains are real and documented. And yet only 4% of those surveyed think human creativity will no longer be needed for building design because of AI. The profession is using the tool aggressively while refusing, almost unanimously, to cede authorship to it. That is not contradiction. That is clarity. The University of Exeter's Ji Han, senior lecturer in design and innovation and co-investigator at the DIGIT Lab, put it precisely in October 2025: large language models can "sometimes reach the level of a novice designer" but remain far behind experts. "The ability to make unexpected connections and analogies," he said, is "still very human." That phrase deserves a moment. Unexpected connections. Not optimised ones. Not statistically probable ones. Unexpected. ## What AI Optimises For Is the Problem AI tends to optimise for what already works, which produces safe but predictable outcomes. This is not an opinion; it is a structural fact about how these systems are trained. They analyse patterns across existing data and generate variations based on what that data contains. They do not experience the gap between what exists and what should exist. Scroll any AI-generated dashboard thread and the result is always the same: rounded cards, soft shadows, neutral grays, indigo or purple accents, two-column layouts with familiar spacing. Technically responsive. Technically accessible. Technically indistinguishable from every other AI output posted that week. This is what optimisation without intention produces. It is competent in the way a form letter is competent. It satisfies requirements without making a decision. Human designers make decisions. More specifically, they make decisions that are informed by things AI cannot access: the memory of a client flinching at a particular shade of blue, the cultural weight of a material in a specific geography, the read on a room that tells you the brief is wrong before anyone has admitted it. As one practitioner noted: AI gets a project 60% of the way in minutes. The remaining 40% is where design actually lives. ## The Clutch Survey, the Budget Lines, and What Businesses Are Actually Buying If the RIBA data covers the supply side, a 2026 Clutch survey covers the demand side, and the numbers align with brutal consistency. According to that survey, 47% of businesses increased their design budgets over the past year despite widespread AI adoption. Only 12% reported a decrease. When selecting designers, 39% of companies ranked creativity as the most important trait, ahead of strategic thinking at 19% and reliability at 17%. Speed and affordability ranked far lower. Businesses are not buying execution. They are buying judgment, taste, and the ability to make a creative decision that holds under pressure. Those are human outputs. McKinsey's 2025 State of AI report confirmed that over 70% of organisations now use AI, but that number says nothing about what they are using it for, or who is still making the call that matters. The split that reveals the real stakes: UX and product design jobs are projected to grow 16% through 2034. Pure execution roles in graphic design grow 2 to 3%. The category being automated is not creativity. It is repetition. ## The Dezeen Survey That the Industry Buried In October 2025, Dezeen published findings from a DIGIT Lab survey that delivered an uncomfortable data point: designers are the creatives most likely to believe AI dulls creativity. Ninety-four percent of creative respondents said they now use AI in some part of their process. Among designers specifically, that figure dropped to 52%. Gen Z data complicated the picture further. Two-thirds of Gen Z creatives use AI tools. But only one in ten believes machine-made work holds genuine creative value. That is the generation that grew up with these tools, not the one defending against them out of professional anxiety. They use the tools and distrust the outputs. That is a more sophisticated position than either the boosters or the luddites are arguing. Saeema Ahmed-Kristensen, director of the DIGIT Lab, said it plainly: "Large language models like ChatGPT and Gemini can generate ideas at speed, remix ideas and surface countless options, but they cannot experience imagination, inspiration or intention." Not won't. Cannot. ## The 40% That AI Cannot Touch Is Not a Niche A March 2026 study from Swansea University, published in ACM Transactions on Interactive Intelligent Systems, found that AI can function as a creative collaborator that encourages exploration, engagement, and inspiration. The key word is collaborator. Not author. Not replacement. The study's framing was careful: the question is not only what AI can do, but how it can help humans think, create, and collaborate more effectively. The research published in Frontiers in Computer Science in 2025 proposed a Human-AI Co-Creative Design Process model and tested it against traditional design process. The result: the hybrid model substantially improved creative performance. For novice designers, AI helped generate more ideas. For experienced designers, it elevated quality and refinement. The variable that determined outcome in both cases was the human designer. The tool amplifies the operator. It does not replace them. Workers with AI skills earn 56% more than peers without them, and senior UX leaders working inside AI-augmented systems now command $160,000 to $190,000. Design skills are, as of 2026, the number one most in-demand skill in AI job postings. Ahead of coding. Ahead of cloud infrastructure. The companies building AI products are hiring humans to make those products worth using. ## The Verdict Is Not Reassuring. It Is a Demand. Human creativity is not being preserved because it is sentimental or because the industry is resistant to change. It is being preserved because it is the only part of the process that makes a deliberate decision about meaning. AI builds from the past. Human designers build toward something that does not yet exist. That distinction is not philosophical. It is functional. It determines whether a building, a product, or a piece of communication does something to the person experiencing it, or simply satisfies a specification. The 96% of architects who rejected the premise that human creativity is expendable were not being defensive. They were being accurate. The prediction: within 36 months, the most valuable credential in any creative discipline will not be software proficiency. It will be the demonstrated ability to make a decision that an AI would not have made, and to explain why that decision was right. Taste, consequence, and intention are the scarce resources now. Every tool that automates execution makes them scarcer, and therefore more valuable. The designers who understand that are already ahead.