Design isn’t output, it’s judgement, and AI doesn’t have any
AI accelerates design execution, but human judgement, creativity and empathy remain essential for meaningful design outcomes.
8 min read
Apr 2026
The acceleration in the use of AI in the design process in recent years has brought into sharp focus the dichotomy between artificial intelligence (AI), as offered by transformer models and human intelligence (HI), which lies at the heart of all design.
This matters intensely because over decades the design profession has learned, often the hard way, that you can only achieve good design outcomes when you obsess about the humans at whom the design is targeted. To help designers avoid getting distracted (typically by bias, trends and ego) the industry has developed a range of principles and processes to help mere mortal designers remember what the actual job is. The best of these centre design processes, decision-making and validation around that human target.
Techniques and concepts such as human factors (the discipline of studying human limitations, capabilities, and behaviours to design systems, tasks, and environments that enhance performance and outcomes) and human-centred design (a problem-solving approach that places real people – users, stakeholders, and communities – at the heart of the design process) provide an aide-mémoire that the best design obsesses over its human target and the worst of design ignores it.
So how does a thoughtful designer reconcile AI’s promise of efficiency gains and creativity aids with HI’s promise of impactful design and meaningful outcomes?
The answer lies in deliberateness and awareness.
Deliberateness – knowing which parts of the process AI can help with (and by corollary which parts of the process AI shouldn’t get close to)
Awareness – understanding the limitations of AI, what its outputs mean (and by corollary what its outputs don’t mean)
Below I outline a personal perspective on the uniquely HI elements of design which AI shouldn’t be trusted with.
1. Identifying and understanding the problem which needs to be solved
AI can do a decent job of doing what it is told, but isn’t close to knowing what it needs to do. For decades now, design professionals have been reinforcing the importance of understanding a problem before designing a solution, with the impact of this evident across some of the most popular digital products on the planet:
Airbnb seeks to solve trust before it solves findability and usability
Xero seeks to solve confidence in self-servicing before it solves small-business accounting
Spotify seeks to solve mood or context before it solves playing your favourite artist
Netflix seeks to solve behaviour-led personalisation before it solves demographic-led personalisation
In each of these cases, product teams used empirical user research (typically drawn from observation) to develop design hypotheses and explore opportunities for competitive advantage. AI is some distance from understanding the relationship between actual human need (observed rather than reported), competitive advantage, consumer expectation and business opportunity.
2. Collaborating, ideating, compromising and working with other people
Design as a discipline has rightly focused on how groups of individuals can best work together to make and implement good decisions. This involves breaking down silos, collaborating well and being creative about ideation and exploration.
Design Thinking includes methods such as how might we, brainwriting and crazy 8s to encourage groups to think creatively. User experience synthesis comprises techniques such as affinity mapping, journey mapping and empathy mapping to help groups of people arrive at a common understanding of user needs and the problem to be solved.
Smart designers have sensitive emotional intelligence and with it an ability to read a room, navigate politics and build trust. A good designer will know when to email, when to raise something in a meeting and when to invite a colleague for a private coffee, all of which will have a positive effect on design outcomes when done consistently well.
There are all sorts of reasons why poor design decisions get made, and the best design processes and cultures proactively mitigate against those. Such mitigations are a thoroughly human endeavour.
3. Interpreting and prioritising findings and determining what bets to make
AI can produce results and make recommendations, but it can’t tell you which results and recommendations matter.
Research is rarely clean or linear because the world is noisy, humans are often contradictory and research findings can be ambiguous. In the midst of that, the UX professional needs to separate signal from noise, which involves spotting patterns, triangulating research and making a best judgement based on the imperfect information available.
Each judgement is a bet. Which feature should we build? Which compromise should I accept and which should I stand firm on? Which edge case is close enough to a use case to deserve a share of project budget? These judgements don’t have right answers, rather they have benefits and trade-offs which are much more elusive.
AI can provide you with options. It can tell you what others did. It can summarise in broad-brush terms what research is telling you. But only humans can decide. Only humans can take action.
The opportunity for eureka moments, product breakthroughs and competitive advantage all lie on the far side of that decision.
Over decades the design profession has learned, often the hard way, that you can only achieve good design outcomes when you obsess about the humans at whom the design is targeted.
4. Applying taste, judgement and creativity to design outputs
I hope I’m not an unthinking or uncritical Apple fanboy, but I do find myself impressed with the articulacy of Messrs Jobs and Ive in interviews, particularly early ones from the 90s which precede them becoming billionaires. One such interview, dated mid-1996, recently appeared on Netflix. At the time Jobs had been fired from Apple (1985), was leading NeXT as it ran out of money (1996), and he (probably) didn’t know it but he was about to return to Apple as CEO (1997) which was similarly running out of money, famously only days from bankruptcy.
Despite his perilous financial situation, he was resolute in his belief in design as a competitive advantage.
The near-penniless Jobs was asked what he felt the difference was between Apple and their juggernaut competitor Microsoft.
His one-word answer was on the money.
Taste.
His answer is important for two reasons. First, it is self-evidently true. Second, in the midst of design frameworks, processes, models, techniques and principles, it is vital to be reminded that there remains space for the expression of that most human of endeavours, creativity.
Modern design methods put guardrails on the design process to connect it to human need, but between the gaps exists the opportunity for creative expression. Taste. Judgement. Distinctiveness. Personality. Differentiation. Delight. Beauty.
There are over 8bn humans on the earth, each of whom has a unique set of skills, beliefs and experiences, expressed through personality and creativity. This has resulted in the creation of design masterpieces from the ceiling in the Sistine Chapel to the Apple Mac, from the Taj Mahal to Google Maps, from the Venus de Milo to the Dyson Airblade.
There are a handful of AI models on earth, each similar to the other. Each is a homogenous representation of everything all of those humans have written and created. Its approach to creativity therefore is to vanilla-fy it to its average.
The opposite of creativity, if you will.
5. Designing a UX process to include the right activities in the right order
In a previous life I owned and ran a user experience agency called Fathom, where I spent the majority of my time selling our services. Like all B2B solutions-based sales, I invested time focusing on the client’s problem (spoken and unspoken) and used that to build a business case for the positive impact of UX. A key part of that was developing a UX process which specifically matched their needs, solved their problem and had the strongest possible business case.
While all recommended processes involved a period of discovery, ideation and solution development, the specifics from project to project were different:
Some clients had good research already completed, others didn’t
Some clients had strong UX maturity, others less so
Some clients had senior stakeholders they wanted to bring on the journey with them
Some clients had development teams who didn’t think UX impacted them and wanted to win over hearts and minds
Some clients had products with good usability but no market validation
Some clients had products which were gaining traction despite being borderline unusable
Some clients were building a new product for a new market segment
Some clients were slowly evolving a legacy behemoth product for internal teams
I quickly learned if I wanted to win work for the agency that I needed to get good at crafting highly impactful project processes, and that the combination of process (what we do and in what order) and methods (what type of research was appropriate at various stages to ideate and validate) was the difference between winning and losing a project.
All of that requires human judgement, sitting in a room with someone, deeply understanding what they are trying to achieve and responding accordingly. In essence, this is the work.
Wisdom is to knowledge what design is to AI.
AI can help with speed, ideation, efficiency, pattern recall and first drafts, essentially the kind of work you might delegate to an entry-level colleague.
However, when the hard work of making judgements starts – problem definition, collaboration, prioritisation, creative decision-making and process development – that’s when human-centred design and human factors thinking need to do the heavy lifting.
