AI and UX design: my experience with what AI and vibe coding are changing in my job

I have always loved UX design because it moves and evolves with its time, its uses, its tools.

I am a curious person, passionate, a bit of a geek too. I like to learn, to question myself, test new ways of working and discover new AI tools (vibe coding).

Not to follow a trend, but to understand what AI and vibe coding really change in our practice as UX Designers.

So naturally, the arrival of tools like Lovable or Figma Make immediately sparked my curiosity.

I never feared them, nor saw them as tools meant to replace me, but rather as new levers to help me in my job: understanding what they make possible, what they shift… and above all, what AI and vibe coding will not replace in UX design.

Observe before you judge (and acknowledge your biases)

When new AI tools arrive, our reactions are rarely neutral.

They are often driven by cognitive biases, sometimes without us even realizing it:

• The status quo bias (“we’ve always done it this way”),
• The negativity bias (“this tool is going to replace me”),
• or, conversely, the innovation/novelty bias (“it’s new, so it must be better”).

As a UX/UI Designer, these are precisely the kinds of reflexes I have tried to avoid.

Rather than judging tools like Lovable or Figma Make too quickly (those famous app builders), I chose a posture familiar to UX Design: observe, test, understand before concluding.

These tools don’t come out of nowhere.

They answer very concrete needs: how to use AI to cut creation time, speed up prototyping, quickly materialize wireframes and mockups, or turn a brief into a first “presentable” version (classic vibe coding).

And on those points, they are effective.

But a limit appears very quickly: they have no context, no empathy, no ability to arbitrate.

They execute from a brief. They don’t challenge the problem and they don’t detect the inconsistencies of a poorly framed need.

Without framing, without user research, without an understanding of usage, the risk is producing inconsistent mockups: it “looks” right, but…

You can quickly end up with choices that create UX debt, hallucinations in terms of ergonomics, and a consistency that falls apart the moment you try to align with a design system (governance, consistency, industrialization).

That’s where I saw very classic biases reappear, but amplified:

the speed bias, the availability bias and the automation bias.

And inevitably, that made me want to go further than the simple “nice tool”: understanding where these solutions complement each other, where they contradict each other, and even starting to ask the question of Lovable vs Figma Make (and, behind it, the question of the “best AI tools” depending on the context).

What these tools actually change in my daily work as a designer

While testing them, I never saw them as replacements.

I see them more as AI assistants for UX design, even as a second brain.

A brain capable of producing faster, suggesting alternatives or improvements based on my first mockups, or generating ideas from a well-defined set of specifications.

Whether with Lovable or Figma Make, you find the same promise: speed up execution, generate variants, move faster from the brief to the prototype.

A brain that is very efficient and very fast at executing, exploring, generating, prototyping… but that has no intuition, no perspective, no fine understanding of context. And even less that ability to arbitrate when the brief is incomplete or politically sensitive within an organization.

Concretely, these tools help me to:

• materialize an idea faster that I have in mind, going from the concept to a first “showable” version (that famous bridge between intention and execution),
• explore several directions without rushing headlong into the first draft, often for lack of time (generate alternatives, compare, iterate),
• free up more time for quantitative research and user research, where the real decisions for the rest of the project are made,
• reduce the “time loss” tied to pure production: mockup variations, prototyping, micro-adjustments.

In the end, they make certain steps of the process easier. They speed up the shaping of ideas.

But they never decide or reason in my place, and that is precisely where the core of my job lies: framing the right hypotheses, challenging a need, detecting inconsistencies, protecting overall consistency (notably with a design system)…

What these tools will not change

If there is one area where these tools quickly show their limits, it is user research.

They can generate interfaces, suggest variants, speed up the shaping of ideas.

But they cannot go into the field.

They cannot listen to a user describe their daily life, nor read between the lines, catch a hesitation, an emotion or something left unsaid. And that is precisely where AI and vibe coding in UX reach their limit: they can execute, but they cannot live the situation, nor understand what is not said.

Research, whether quantitative or qualitative, rests above all on a deeply human understanding of usage. On the ability to connect numbers to real behaviors. To turn verbatims, sometimes contradictory, into actionable user needs.

That is exactly where my role takes on its full meaning:

• Interpreting data,
• Cross-referencing weak signals,
• Surfacing problems that are not always explicitly stated,
• And translating all of this into design decisions.

Ultimately, the more powerful AI Design tools become, the more valuable this phase becomes.

Because without empathy, without framing and without solid research, all you do is go fast, but not necessarily in the right direction…

In the end

Design has never been a frozen profession.

Vibe coding solutions don’t redefine what the UX/UI Designer profession is. They simply reveal its essence.

When tools generate faster, thinking becomes more valuable.

And when technology advances, the human role (exchange, empathy, understanding and decision) becomes central.

These tools are neither threats nor ready-made answers.

They are, as I mentioned earlier, facilitators and accelerators. But they replace neither user research, nor UX framing, nor the arbitrations.

You just have to tame them and know how to use them in the right place, at the right time, and with the right perspective.

My job is changing. As always. And it will keep changing.

And that is precisely why I keep practicing it with passion, curiosity, exigence… and enthusiasm!