Ambient Design: AI Does Not Replace Human Judgment

Taste and Judgment: A Challenge for AI

In a context where ambient design is increasingly utilized, Itamar Medeiros highlights a crucial aspect: taste has become the new bottleneck. He asserts that what distinguishes good work today is not so much the ability to generate ideas, but rather the ability to effectively discriminate between them. With a range of plausible options available, the true skill lies in choosing which ones should be highlighted. During a live panel on the state of AI in UX and product design, Christian Eckels, product designer at CNN, expressed a strong opinion on this subject. According to him, “AI can polish a bad UX; the judgment, taste, and responsibility lie with the designer.” This statement underscores that, while AI can enhance certain technical aspects, it cannot replace human judgment.

Taste is a quality that is difficult to define in words. Designers may have a clear idea of what they want, but when they use AI tools, the result can often seem generic or disconnected from their personal vision, even if it technically meets the initial brief. Taste is not an innate quality; it develops through repetition, critique, and exposure to excellent work over time. Most experienced designers have refined their taste by examining their past work and seeking elegant and creative solutions to improve it. While designers are not the only ones with taste or judgment, they have the professional responsibility to exercise it, maintain a high level of quality, identify what is lacking, and advocate for the interests of the end users who will live with the decisions made.

The Importance of Unique Words

In the realm of ambient design, UX microcopy is often standardized, filled with phrases like “Continue your journey,” “Let’s go,” or “Almost there.” While these phrases are functional and direct, they lack memorability. What makes a brand unforgettable lies in the choice of words. Current ambient design tools fail to reproduce this uniqueness. You are the only one who understands the soul of your product and what makes it special.

Design and product teams have valuable resources in old tone documents, written long before the advent of ambient design tools. These guides capture the nuances that make a brand recognizable, from the cadence and structure of sentences to communication principles. They also define which words to use to ensure the user experience remains true to the brand's identity.

Design Systems and Their Derivatives

Ambient design struggles to maintain consistent design systems. While these tools can be connected to your tokens and components, like Figma Make drawing from your Figma library, or Lovable being configurable with custom themes, the generated results tend to drift. Most ambient design tools rely on AI models, often LLMs or multimodal models. These models, being probabilistic, produce slightly different results with each request, leading to gradual drift and erosion of the design system from within.

This drift is one of many challenges these tools can pose. A study titled Qualitative Evaluation of GUIs Designed by LLM tested three leading models (GPT o3-mini-high, DeepSeek R1, and Claude 3.5 Sonnet) by generating UI mockups for a chat system, a technical team panel, and a management dashboard. Experts concluded that while the models are effective at producing structured layouts, they struggle with accessibility standards and providing interactive features. Accessibility standards and interactive states are essential elements that any design system specifies, including contrast ratios, focus indicators, keyboard management, and component states, and AI models do not adhere to them by default.

So, does the ambient design tool you are using strictly adhere to the specifications of your design system? The answer is probably no.

Understanding the End User

The ambient design tools you are considering using are often based on models that have a generic view of the user, but not of your specific end user. Designers and UX researchers spend the majority of their time identifying who the user is and what they need, a task that cannot be delegated to a simple prompt or AI-generated “vibes.”

According to the report The Future of User Research 2026 by Maze, 69% of researchers now use AI in their work. These researchers leverage AI to manage repetitive execution tasks while focusing their energy on aspects that only a human can handle: formulating more precise questions and interpreting nuances. An AI model cannot account for what is not present in its training data. For example, a user with an unusual workflow remains invisible to the ambient design tool. This type of situation only appears in AI results when a designer or researcher presents the work to a real user and observes the interactions.

The Irreplaceable Role of Designers

The barrier to producing something that looks ready to ship has significantly decreased. Today, anyone can generate a user interface, but does that mean they can actually design a product? Ambient design empowers non-designers to produce, but the discipline of design itself is more complex to automate. This discipline begins after the first output generated by the tool. It involves choosing a specific layout from several options or deciding why a certain hierarchy should be used for a particular user at a specific moment in the user journey.

A screen designed with ambient design often resembles what was once the starting point of the design process: an interface direction before the difficult decisions behind it have been tested and resolved. The designer's job is to evolve this starting point through the compromises that arise once a product meets reality, such as defending decisions to the engineering team and adapting flows based on user feedback. AI jumps straight to the final artifact, but in this leap, the reasoning process that transforms an interface into a product decision is lost. Ambient design merely compresses execution.

Documentation and Understanding

When a designer builds something manually, the work produces traces. In other words, the work generates documentation as a byproduct of the process. In contrast, when ambient design generates an output in a single leap, many intermediate layers disappear. The artifact exists, but its provenance is incomplete or fragile.

The notion of "understanding debt" can be applied here: the growing gap between the volume of work that exists and the volume that any team member understands. Imagine a situation where, six months after shipping a flow designed with ambient design, someone from your team asks why the cancellation flow has three confirmation steps, while the upgrade flow has only one. The designer who built it may not remember, as the reasoning was not recorded or highlighted by the tool. The interface exists, but the reasoning behind it may not be documented.

Ambient design accelerates production faster than teams can preserve understanding.

Training the Next Generation of Designers

Junior designers learn by doing the hard, sometimes “ugly” work. They may go through fifteen versions of a layout, receive feedback, and gradually develop the eye that becomes the foundation of taste and judgment. This work is the curriculum. There is no other way to develop judgment, which is irreplaceable and belongs to designers.

The repetitive work that builds judgment is the learning that produces the next batch of senior designers. If there is a tool that removes steps from the learning cycle, seniors and design leads must reintroduce it through critique, explanations of trade-offs, slowed reviews, and making decision-making visible. Not everything should be accelerated.

Ambient design has its place in a senior workflow, but the path to seniority involves repeated practice and reflection.

Intentional Ambient Design

This article is not an anti-ambient design manifesto. AI makes me a better designer and leader, and I am excited (as well as intrigued) by what this new wave will bring to the industry. The tools I have tried are excellent, and they have reminded me why I fell in love with design in the first place. So, while fully embracing ambient design, we should be careful not to confuse the generated vibes with the art of design itself.

The lasting value of design lies in judgment, systemic thinking, language, research, responsibility, and the ability to translate ambiguity into coherent experiences. The teams that will thrive in the era of ambient design will not be those that resist AI. They will be the ones that use it deliberately while protecting the human capabilities that make products meaningful in the first place.

Arin Bhowmick (@arinbhowmick) is Chief Design Officer at SAP, based in San Francisco, California. The above article is personal and does not necessarily represent the positions, strategies, or opinions of SAP.