What AI Still Can’t Do: The Design Skills That Matter More Than Ever
This page may contain links from our sponsors. Here’s how we make money.
Every few weeks, a new AI tool launches with claims about revolutionizing design. The demos look impressive. The generated outputs appear polished. And designers everywhere feel that familiar knot in their stomachs.
But here’s what the product demos never show you: the messy, human parts of design that happen before and after the pixels. The parts that determine whether a project actually succeeds or quietly fails.
According to the 2025 State of AI in Design report from Foundation Capital and Designer Fund, 89% of designers say AI has improved their workflow in some way. Yet dig into the Figma 2025 AI Report and you’ll find a telling gap: while 82% of developers are satisfied with AI output, only 54% of designers say AI improves the quality of their work. That 28-point difference reveals something about where AI actually helps and where it falls short.
Asking Better Questions
Before any design work begins, someone has to figure out what problem actually needs solving. This sounds obvious until you realize how often projects fail because they answered the wrong question beautifully.
AI tools excel at producing outputs. Give them a prompt, and they’ll generate endless variations. But they cannot interrogate whether the prompt itself makes sense. They cannot push back on a brief that misunderstands the user. They cannot recognize when a client is asking for a landing page redesign when they actually need a positioning strategy.
The Figma report found that only 31% of designers use AI for core design work like asset generation. The other 69% are using it for peripheral tasks. Why? Because the diagnostic work of understanding what to make still requires a human in the room, listening to what stakeholders say they want and identifying what they actually need.
Navigating Organizational Reality
Last month I watched an AI tool generate a beautiful homepage concept in thirty seconds. Then I watched a designer spend three weeks navigating the approval process: aligning the marketing team’s conversion goals with the brand team’s consistency concerns, while managing an executive who wanted his favorite color incorporated somewhere prominent.
No AI can do this work. It requires understanding that the VP of Sales needs to feel heard even when his suggestions won’t be implemented. It requires knowing when to push back on feedback and when to let it go. It requires building relationships over time so that when you advocate for users, people actually listen.
According to Resume Genius’s 2025 hiring survey, 35% of hiring managers have encountered AI-created portfolio projects. But the same survey shows that interpersonal skills and demonstrated collaboration remain among the top factors that actually get candidates hired. When everyone has access to the same generative tools, your ability to guide human decision-making determines whether good design actually ships.
Systems Thinking at Scale
AI can generate a gorgeous button. It struggles to generate a design system that scales across twelve products, accommodates WCAG 2.2 requirements, anticipates future feature additions, and remains maintainable by a rotating development team over three years.
This kind of thinking requires holding multiple timeframes in mind simultaneously. What works today? What breaks next quarter? What becomes technical debt in two years? AI optimizes for immediate outputs. Humans plan for sustainable outcomes.
The same limitation appears in service design and experience mapping. AI can help visualize individual touchpoints, but understanding how those touchpoints connect across channels, how they feel different for different user segments, and where the gaps create friction requires synthesis that current tools simply can’t perform.
Ethical Judgment
AI tools optimize for engagement metrics and aesthetic patterns. They don’t ask whether a design is manipulative. They don’t consider whether a dark pattern might increase conversions while eroding trust. They don’t weigh the impact of design decisions on vulnerable users.
This matters more now precisely because AI makes certain approaches easier to implement. When you can generate infinite variations of a checkout flow, someone needs to decide which variations respect users and which exploit them. That judgment requires values, not just pattern recognition.
The Figma report found that 80% of both designers and developers believe learning to work with AI will be essential to success in their role. But the report also notes that AI projects still lack clarity in purpose: only 9% of teams name revenue growth as the top goal, while 76% cite vague aims like “experimenting with AI.” Someone has to translate those experiments into coherent strategy.
The Taste Gap
Perhaps the most underrated human skill is taste: the ability to look at twenty AI-generated options and know which one is right. Not which one is most polished or most on-trend, but which one actually serves the project’s deeper goals.
Taste is accumulated judgment. It comes from years of seeing what works and what doesn’t, from studying design history, from understanding why certain choices feel timeless while others feel dated within months. AI can mimic existing patterns, but it cannot develop the contextual sensitivity that distinguishes appropriate from generic.
The 2025 State of AI in Design report describes AI output as “good enough but not perfect.” That gap between good enough and right is where human designers earn their value.
Where to Invest Your Energy
If AI handles more execution, invest more in what AI cannot replicate. Get better at facilitating conversations that surface real requirements. Develop your ability to manage stakeholder dynamics. Practice thinking in systems rather than screens. Build your ethical framework for evaluating design decisions. Cultivate taste through deliberate study of what makes design work.
According to the St. Louis Fed’s 2025 research on generative AI adoption, the share of work hours spent using AI increased from 4.1% to 5.7% over nine months. The technology is becoming embedded in daily work. But notice what that number also tells you: 94% of work time still involves tasks where AI isn’t the primary tool.
The designers who will succeed aren’t those who become the best prompt engineers. They’re the ones who become indispensable for everything that happens around the prompts.
