Whether we want to admit it or not, we all use artificial intelligence to change our curtains, find recipes, plan a trip to another country, and, yes, even for work. If you’re reading this, I’m pretty sure you’ve used it too. UX/UI design and graphic design haven’t been left out either. I wouldn’t say AI alone is the reason the job market in these fields reached one of its most difficult points between 2023 and 2025.
Imagine, today we can write down an idea and, within minutes, get a website, a logo, a flyer, or a social media design. I imagine some companies may have asked themselves, “Why would I need a UX/UI or graphic designer if I can generate it with AI?” But by providing very general instructions, they may also have delegated important business decisions to the tool, resulting in designs that differ from others only in their name and perhaps their color.
I wonder if the same ease of generating designs that may have led some companies to question the need to hire designers is also helping them recognize their value. Once again, I’m not attributing these changes entirely to AI, but I do think we can ask ourselves what value design judgment brings at a time when generating proposals is becoming easier.
For me, part of the answer is understanding the purpose, giving the proposal an identity, and making sure it works for the people who will use it.
AI can help us create good designs. But to get a more personalized result, we need to give it direction.
Here are some of my tips for creating designs with AI.
1. Define the purpose before designing
Before choosing colors or typefaces, ask yourself:
What am I creating, and who is it for?
What do I want that person to understand or do?
Where will they see or use the design?
An application for viewing medical results and a website for selling clothes have different priorities. The same goes for a workshop flyer and an advertisement for a coffee shop. The purpose should guide your visual decisions.
2. Give it a clear identity
Words like beautiful, modern, or professional are a starting point, but they can mean different things to different people.
Describe the brand’s personality, colors, typography, image style, and tone. For example, instead of asking for “a modern design for a coffee shop,” you could say:
“I want a warm identity for a neighborhood coffee shop. We’ll use cream, brown, and olive green tones, real photographs of the products, and a typeface with personality for the headings. The design should feel welcoming and easy to read.”
The clearer the direction, the easier it will be to evaluate whether the proposal represents what you’re looking for.
3. Share references and create rules
References help communicate an idea, but explain what you like about each one: its composition, use of space, photography, or the way it highlights information. You can also show what you don’t want.
Look for inspiration in magazines, packaging, posters, photographs, or physical spaces, in addition to other websites.
Once you find a direction that works, document your decisions: colors, typography, logo usage, images, and the tone of your text. If you’re creating an application, add rules for buttons, forms, tables, modals, and messages.
These rules define not only how components look, but also how they behave: what happens when a button is loading, how an error is displayed, or how table filters work.
This gives you a guide for maintaining consistency across your future designs or screens.
4. Review the fundamentals and real-world use
A design can look attractive and still fail to fulfill its purpose. Review its hierarchy, contrast, alignment, spacing, consistency, and readability.
Then, look at it in the context where it will be used:
Application or website: Does it work on mobile? Can it be navigated with a keyboard? What happens if something is loading, fails, or has no data?
Flyer or social media post: Is the message clear and readable on a phone?
Brochure or printed material: Is the text large enough to read, and is the information well organized on each side?
Always review names, dates, links, and details before publishing or printing.
5. Give specific feedback
The first proposal doesn’t have to be the final one. Explain what works, what doesn’t, and why.
Instead of saying, “I don’t like it; make it prettier,” you could say:
“The title is competing with too many elements. I want it to be the first thing people read, followed by the date and location. Reduce the decorative elements and add more space between sections.”
You can also ask for several proposals with different compositions and compare which one communicates your idea best.
Tools you can explore
Not every tool solves the same problem. Some help define an identity; others help create visual materials or study interface patterns.
Explore colors: Coolors and Adobe Color
Find typefaces: Google Fonts
Create flyers, social media posts, and presentations: Canva and Adobe Express
Explore interfaces with AI: UX Pilot, Google Stitch, and Figma Make
Study interface components and patterns: Material Design, IBM Carbon, AWS Cloudscape, and Atlassian Design System
Design systems are especially useful for learning how other organizations solve common problems. The idea is to study their decisions and adapt what makes sense for our product.
A prompt you can adapt
I want to create [type of design] for [name or description of the project].
The goal is to [what I want to communicate or what action I want the person to take].
It is intended for [audience] and will be used on [Instagram, a website, in print, etc.].
The brand’s personality is [characteristics]. Use [available colors, typefaces,
photographs, illustrations, or references]. The most important information is
[main content]. Organize it so that it is easy to identify and read. I want to avoid
[styles or elements that don’t fit the brand]. Generate three proposals with different
compositions and briefly explain the decisions behind each one. Then help me review
which one communicates the message best and what I should validate before publishing it.I enjoy using AI because it helps me explore possibilities and move forward. But the easier it becomes to generate a proposal, the more important it is to know how to observe it, question it, and give it direction.
The tool can generate options. We decide what we want to communicate and how we want it to feel.
Sources
Indeed Design (2023): UX Job Listings Plunged in 2023 — data on the decline in UX job listings on Indeed.
Nielsen Norman Group (2026): State of UX 2026: Design Deeper to Differentiate — an analysis of the UX job market and its signs of stabilization.
Microsoft Research (2026): Interrogating Design Homogenization in Web Vibe Coding — explores the risk of AI-generated web designs looking alike.
Federal Reserve Bank of New York (2026): Do Job Postings Show Early Labor-Market Effects of AI? — an analysis of the relationship between AI and job postings.