Use AI to redesign my student capstone project

Recently, I started integrating AI into my daily design workflow, and it has genuinely challenged my understanding of the traditional design process.

To test its limits, I rebuilt my university capstone project using AI. I fed my original design system, problem scope, and user task flows into Gemini and Lovable. The outcome was absolutely amazing. AI generated a high-fidelity prototype with a polished UI, great UX copy, and functional audio features (**Click the volume button, and the prototype actually pronounces the dish’s name).

But does this prototype actually solve the problem?
“I felt like the information was overwhelming... and the design feels too AI.” — This was the feedback from a friend who tested it.

I love this critique because it forced me to ask "WHY."

- How do you define an “AI-vibe” design? Why do people perceive AI-generated work as "cheaper" or less thoughtful, even when human design is rarely perfect? There are many clients asked for design works that has no “AI-vibe”. They pointed out they will not pay for things looks very “AI”.

- Why can a flawless, over-engineered product sometimes trigger user resistance?

- In an era where everyone claims the GUI is becoming obsolete, why do users still care if an interface looks AI-generated? If the UI still matters, where exactly does great, human-crafted design deliver its true value?

Right now, I view AI primarily as a powerful design collaborator. For zero-to-one or small product features, my workflow looks like this:

Human: Define the problem scope and core design system.
Human: Map out the MVP task flows.
AI: Generate the initial high-fidelity prototype.
Human: Simplify, refine, and iterate on the prototype.
Human: Conduct usability testing → gather feedback → determine next steps.

I'd love to hear from other designers—
How do you use “AI” in your design? 
Where do you think exactly human-crafted design deliver its true value?