As generative AI tools have become embedded in design workflows, they've delivered on their promise of speed and efficiency. A designer can spin up a landing page, a product interface, or an entire app prototype in minutes instead of days or weeks. The output often looks impressive on the surface but flaws often find their way to the top and designers should be vigilant about detecting and ultimately avoiding them. Using AI responsibly allows for ideation and accelerated execution. Simultaneously, users must preserve the human judgment required to create something meaningful and intentional. Let’s dive into how using AI tools can go right and oftentimes very wrong.

The allure of the AI-powered design tool

Imagine a designer’s first time using an AI tool to generate a design. Tools like Claude Design, v0, and ChatGPT are incredible for creating mockups and prototypes. The designer enters her first prompt and is astonished by a clean UI that appears to handle the problem effectively. However, this initiation can be a gateway. As the designer is distracted by speed and efficiency, she may find herself missing critical design flaws.

Designers should keep in mind that AI tools may put all the right pieces on the chessboard, but it may not put them in the right places. Perhaps our designer didn’t notice that the save button on a form wasn’t in its usual spot. Maybe account management is too buried in the navigation for the user to easily find.

Considering the blueprint of a house can paint a helpful picture. An AI tool may lay out the kitchen with the dishwasher too far away from the sink. The specs require a dishwasher and a sink to be in the kitchen, and the AI tool technically handled that. An architect, however, would take into consideration that someone loading the dishwasher would not want to walk back and forth from the sink to the dishwasher since that would naturally be a poor user experience; ideally, the two would be placed next to each other. But this is the kind of oversight that an AI tool may miss, and it stresses the importance of ensuring that the architect stays close to his craft.

When AI takes away agency

Returning to our designer, she appears to be solving problems but is slowly prompting away her design decisions. The tool begins to gain more authority over the final product. For some, the takeover starts slowly; for others, it quickly becomes an addiction. Before she realizes it, the designer has promoted the tool to creative director. This scenario illustrates a key point: it’s far too easy to rely on an AI tool to do the thinking the designer should be doing.

The loss of the unique designer

Unchecked AI tools can produce output that looks like everything else on the internet. Social media ads and event flyers often share the same palettes, fonts, and animation styles. Similar to how many image generation tools churn out cut-and-paste designs, AI-generated designs often output cookie-cutter templates that result in generic, flavorless experiences.

One of the best parts about a great UX designer is their unique, creative eye. The designer’s ability to see the nuance in the user experience holds immense value. Designers can now plug their design systems directly into these AI tools, but there is still a watermark that can leave its trail on designs.

There’s a parallel in software engineering through a concept called “code smell,” which means that a particular area of the code feels off. The code repeats itself unnecessarily, there is a lot of code bloat, or there are an unreasonable number of bugs. Whatever the reason, no developer wants their work to have a bad code smell. Similarly, I’d argue that designs can have a bad “design smell,” and AI tools can contribute to that. I might even push the boundary on that concept and say that AI-generated designs can have a bad “AI smell”: a faint yet unmissable stench of something not being authentically human, organic, and sound. It might surface in a color being slightly off, a button being out of place, or be far more obvious. Designers should take care to ensure that their work, whether AI-enhanced or not, avoids this so-called “stench.”

The problem was never the AI tool

The pros and cons that AI tools provide is the double-edged sword the design industry isn't talking about loudly enough. We've optimized for output volume at the cost of design integrity and psychology: the deeply human science of understanding how a user moves through an experience, what creates friction, and what fosters trust. Designers should remember that quantity is not a valid substitute for quality, nor are AI-generated designs meant to be a true replacement for a designer’s original concepts. A balance is needed to ensure that designers are staying true to both themselves and their users, relying on AI tools only to assist, not replace. There is nothing wrong with using AI tools to accelerate productivity, but designers (and arguably any user of AI tools) should remember that AI is meant to supplement one’s effort, not replace it.

Prompting must also become a language in which designers are fluent. A tool’s output is only as good as the user’s input, so keeping things descriptive and clear is paramount to a successful result. In addition, whenever designers use AI, they should keep a watchful eye for anything that may be off in the design. Lastly, designers shouldn’t forget that there is no replacement for their intuition. They should continue to rely on their design skills, understanding user goals, and psychology to remain successful (and, dare I say, relevant). Designers need to remain highly involved in the design process, viewing AI tools as assistants – or perhaps interns – who still need constant supervision and guidance. By exhibiting restraint and treating AI as a companion for ideation and execution, designers can define the next era of software without losing the human element that makes good design matter.