UX designers, you can breathe a sigh of relief: Any company that thinks it can replace its UX designers with artificial intelligence (AI) tools will have to think twice.
First, there are simply too many parts of a UX designer’s role that AI isn’t even close to understanding or performing for it to be either a realistic option or a sensible strategy right now.
Second, UX design is about designing for humans. Naturally, to design for humans you have to understand humans, and this is something AI is not capable of in any meaningful way.
There’s a but, though, and it’s a big one. There are some things that a UX designer does that AI can replicate to some extent.
And when you throw into the mix the fact that some companies are always looking for ways to cut costs and the fact that the capabilities of AI are advancing at breakneck speed, you can see why some people imagine a future in which the craft of UX design is under pressure from AI. But the evidence really does not support that conclusion, thankfully.
We’ve written this blog—without the assistance of ChatGPT—to give you an overview of the impact of AI on UX design.
We’ll first investigate whether AI can replace UX design. We’ll do this by going through the broader responsibilities of UX designers one by one and seeing whether they can be AI-automated.
Then we’ll look at three ways you can use AI to your advantage as a UX designer and take a wider look at the future relationship between UX and AI.
Finally, we’ll wrap things up with some key takeaways.
As always, if you want to skip ahead to a certain section, use this clickable menu:
- The impact of AI on UX design
- Can AI replace UX designers?
- How to use AI in UX design
- What is the future of AI in UX?
- A closing thought
1. The impact of AI on UX design
There’s been a lot of noise about the impact of AI on UX design.
One thing we do know is that UX designers are involved in building AI. Meta is currently hiring for their Generative AI team and until fairly recently Open AI was looking for product designers.
But to what extent is AI involved in the UX design process? What AI tools are we talking about?
There are already hundreds of AI tools out there that could potentially help with UX design, but let’s quickly summarize seven of the main ones:
- ChatGPT: Generates conversational outputs, research, copy, and more based on your prompts.
- UIzard: Scans your sketches and turns them into editable, testable digital designs.
- Copy.ai: Creates different types of copy based on being prompted with the brand, product name, and description.
- Jasper.ai: Generates custom copy so you can avoid lorem ipsum, similar to copy.ai in some ways.
- Fontjoy: Speeds up the process of font pairing.
- Khroma: Creates infinite color palettes according to your prompts.
- Midjourney: Generates images from natural language descriptions (prompts), and can also be used for generating icons, mood boards, and more.
Check out our full guide to the best AI tools for UX design here.
In this blog, we’re going to focus on the impact ChatGPT and its design plugin, FigmaAI, have had on UX design. In the next section, we’ll explore whether AI can replace UX designers, by diving into what ChatGPT and its FigmaAI plugin can and can’t do.
(For a balanced overview of AI’s impact on tech more broadly, our blog Big Tech Layoffs 2023: Are Tech Jobs Secure? is a great place to start.)
2. Can AI replace UX designers?
A lot of UX design is about empathy of course, but we can look at it in an even more basic way—a lot of UX design is about talking.
So, to answer the question of whether AI can replace UX designers, let’s experiment with another question:
Can UX designers do their job without talking to people?
Or, we can be more specific:
Q: What parts of a UX designer’s job involve talking to people—either in-person, remotely, or via email or some other form of communication (email, Slack, JIRA, Asana, etc)?
A: A lot, is the simpler answer. Those of you who are UX designers, who are on your way to becoming one, or who work closely with them will already know this, of course.
But let’s break it down and answer the question in a formulaic way. This table shows 15 duties that a UX designer would typically carry out—over a period of a few months, let’s say—and how often or not they involve talking to people (with options of: Always, Often, Sometimes, Rarely, Never):
Duty: Collaborate with UX research to create user personas
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Collaborate with UX research to understand pain points
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Run a simple internal user test or a wider external one
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Understand which UI components fit which context
Does it involve talking to people? Often
Can it be replaced by AI? No
Duty: Build prototypes in Figma (or a similar platform)
Does it involve talking to people? Often
Can it be replaced by AI? No
Duty: Ask PM or engineers for clarification on requirements
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Join in a design crit, either as a moderator, presenter, or critiquer
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Plan a sprint
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Prioritize tasks
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Run or participate in a design sprint
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Present and explain work and process in a sprint review
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Present work and explain thinking behind decisions to stakeholders
Does it involve talking to people? Always
Can it be replaced by AI? No
Duty: Do competitor research
Does it involve talking to people? Rarely
Can it be replaced by AI? No
Duty: Create implementation guidelines or documentation
Does it involve talking to people? Often
Can it be replaced by AI? No, although ChatGPT could help to an extent in the writing process
As you can see from this table, 13 of these 14 tasks that a UX designer would typically have to perform involve regularly talking to people, and only one—doing competitor research—doesn’t.
It might sound simplistic, but until AI can engage in conversation like an intelligent human—asking and answering questions, actively listening, showing empathy, understanding connotation and subtext—it can’t really do any of the things a UX designer does.
Because AI can’t do anything that involves conversing with other people, it can’t do UX design—or any other job that requires interpersonal communication.
But, before we move on, let’s quickly look at a real and current job description at Meta for a Product Designer in their Reality Labs. Could AI perform any of the responsibilities their Product Design will be expected to carry out?
Product Designer: Reality Labs at Meta
Responsibility: Take broad, conceptual ideas and turn them into something useful and valuable for people using our products
Can this be done by AI? No
Responsibility: Work collaboratively with Product, Research, Engineering, and other partners from concept to completion
Can this be done by AI? No
Responsibility: Drive planning, strategy, and vision on key priorities across cross-functional teams
Can this be done by AI? No
Responsibility: Own the end-to-end design (user flows, interaction, visual, etc.) across multiple products and projects
Can this be done by AI? No
Responsibility: Present designs, prototypes, and concepts to cross-functional partners and stakeholders
Can this be done by AI? No
Responsibility: Provide implementation guidance to engineers and ensure features launch at the highest quality
Can this be done by AI? No
Responsibility: Clearly articulate design decisions to a vast set of internal stakeholders across Reality Labs and Meta
Can this be done by AI? No
This example again shows just how far AI is from replacing UX or Product Designers. All of the key responsibilities can only be carried out by an actual human.
But, circling back to the first table which included 15 UX designer responsibilities, there was one—carrying out competitor research (if not understanding or gaining insights from it)–that is doable without talking to people.
Is this something that AI can replace UX designers doing? Or at least something UX designers can delegate to AI?
Can AI carry out competitor research as well as a UX designer?
When it comes to carrying out competitor research, the one duty that may be achievable without talking to people, ChatGPT is an AI that could be somewhat helpful.
Even though it wouldn’t be able to carry out research to guide or influence your own design decisions, it could possibly help with identifying competitors (although as a UX designer, you’d expect to already be familiar with the competitors) and creating basic written summaries of their key user journeys.
Let’s see what happened when we asked it to do just that.
We played the role of a UX designer at Headspace (or a similar meditation app) and tried to conduct some basic competitor research.
First, we asked ChatGPT to describe Aura’s user journey for canceling a subscription and this is what we got back:
ChatGPT struggles to give anything beyond a basic, vague description of a user journey.
ChatGPT begins its output with:
“Please note that as an AI language model, I can provide a generalized user journey for canceling a subscription with Aura, but specific steps and processes may vary.”
And, true to its word, it quickly becomes apparent that the ChatGPT-generated user journey is not from Aura, but rather a summary of data/content on common journeys for canceling a subscription.
As you can see ChatGPT is unable to determine if Aura has an “Account” or “Profile” option (step 2), or a “Subscription” or “Billing” tab (step 3).
It also includes a lot of conditional verbs, such as “may” (step 5), “might” (step 6), and “should” (step 7), to emphasize that this is a hypothetical user journey, not an actual one.
Because this is a hypothetical and vague description of the user journey, it would be worthless as competitor research for a UX designer.
Even so, we wanted to see if ChatGPT could identify potential pain points in Aura’s cancel a subscription user journey:
ChatGPT can’t describe pain points users may encounter in Aura’s cancel a subscription user journey
Similar to its output for the previous prompt, ChatGPT is unable to give anything beyond vague, hypothetical pain points for canceling a subscription user journey.
Again its output is punctuated with “if”s and “might”s, which renders it totally unusable for the purposes of competitor research.
For competitor research to influence, guide, or inform UX design—whether that’s on what to do or what not to do—it needs to be clearly visualized and include at least some attempt to uncover the deeper thinking and highlight what would be more and less likely to work.
ChatGPT is simply not able to do that at this stage in its development.
So, if AI can’t help us with competitor research, what parts of UX design can it actually do?
Next, we’ll look at a couple of examples of how you might actually be able to use AI to support UX design.
(If you’re looking for an analysis of AI’s impact on Web Development, AI and Web Development: A Coder’s Perspective in 2023 is a great starting point.)
3. How to use AI in UX design
We started here by looking at the functionality of Chat GPT’s FigmaAI to see what the plug-in for the UX designer’s best buddy could offer.
Right now, the main things FigmaAI can help with are creating and then styling and/or editing multiple images and creating texts.
Creating and styling an image
Instead of having to source images from libraries and other means yourself you can automate and use FigmaAI to speed up the process, if not fully automate it.
There’s a three-step process to do so:
- Enter a basic prompt of what you want to see (“A view of a city at night from above” is the example used in the website demo of FigmaAI.
- Choose a style if you want one (the options are Realistic, Draw, Artistic, Picture, and Abstract)
- Generate the image and then edit it and generate more until you have what you need
FigmaAI, ChatGPT’s Figma plugin, can generate and edit multiple images within the UI
By using FigmaAI to generate and edit images you might save quite a bit of time compared with sourcing them and editing them yourself, although it remains to be seen if the quality is always up to the same level.
Creating text for your design
FigmaAI promises to “Create better text for you design” (ignoring the typo in the headline, not a great look for an AI text generator), again following a straightforward three-step process:
- Enter a basic prompt of what you want the text to say (“Headline for a sustainability-focused travel landing page” is their example).
- Select the level of creativity and formality you want for the tone of the text.
- Generate the text and then use prompts to change the length or tone.
FigmaAI can generate, translate, and edit texts of varying quality
The headline it generated “Experience the World While Making a Positive Impact on the Environment”, while bland and sounding exactly like you’d expect from an AI text generator, is probably just about ok as a starting point for the actual headline.
With some tweaking here and there it could turn into something strong enough to be user-facing. The question then would be whether writing the prompt was actually quicker than drafting a headline of the same quality.
That would—of course—depend on the writing abilities of the UX designer, but it’s clear this tool will add value for some UX designers, especially those who don’t have UX writing support or are writing in their second or third language.
For some more perspective, we decided to see what ChatGPT offered us for the headline with a very similar prompt, “Create three headline options for a sustainability-focused travel landing page.”
ChatGPT offers better headlines than FigmaAI with the same prompt
Interestingly, although not perfect, the three options ChatGPT generated were all stronger and more natural sounding than FigmaAI’s effort. Here’s what it churned out:
- “Embark on an Eco-Adventure: Discover Sustainable Travel Experiences”
- “Explore the World Responsibly: Unforgettable Journeys with a Sustainable Twist”
- “Travel with Purpose: Embrace Sustainable Tourism for a Greener Planet”
This suggests that—for now at least—ChatGPT might be more useful to UX designers who are looking for AI support with creating texts.
In this video, product design professional Maureen Herben looks at some more ways AI can be applied in UX:
Now that we’ve looked at a few examples of how you can use AI in UX design, let’s think about the future of AI in UX.
4. What is the future of AI in UX?
Sticking with ChatGPT, although it’s not able to help in any meaningful way with carrying out user research or competitor research, it is able to make the text more concise at the same time as removing typos and grammatical errors.
This could open the door for UX designers to use it for a few more things:
- Editing and improving documentation or implementation guidelines.
- Editing and improving content for presentations (but it would have to be inputted in text format, rather than on slides or similar).
- Creating basic marketing content.
And it’s easy to imagine that as ChatGPT evolves, it would eventually be able to do these three things at higher and higher levels. (The same applies to the FigmaAI capabilities we looked through earlier.)
It is also useful to an extent for generating ideas. Let’s imagine we’re a UX designer creating an app to rate and review independent coffee shops in Berlin and we prompt ChatGPT with:
“I’m creating an app to rate and review independent coffee shops in Berlin. What features would my users expect to find in my app?”
Here’s what we get back as output:
And so on. Although these are all very basic ideas that most UX designers would already know and the rest would be able to Google in a few minutes, the list could still be at least somewhat useful as a starting point.
What might be more interesting would be if as the same UX designer we were involved in the app naming process and asked ChatGPT for ideas:
The 10 name suggestions we get back from ChatGPT range from the odd (JoeReview) to the pretty good (BrewBerlin, BeanScout, BrewReview).
And maybe that gives us the clearest indication of what the future of AI in UX is. Although AI won’t be able to replace UX designers in the foreseeable future, it can be a useful tool to help them with idea generation, writing, editing, and naming.
Combined with the other tools we touched on in the first section, becoming UX designers who are familiar and confident with these tools can start to give themselves an advantage.
5. A closing thought
There’s a quote about the nature of UX design, from Erica Ellis, Sr Manager, Head of Product Equity Design at Uber:
“We are in the business of solving problems to meet the needs of the people who are going to use our product. That’s the driving force in everything we do. The people. We design for people… Every decision must serve a greater purpose. We create products that people can and want to use — products that will make their lives easier, and better. To create that experience, to design for people, that is what I do.” (We’ve added the bolding.)
To design for people you need to understand people, and to understand people in any meaningful way you need to talk to them and you need to listen to them.
A world in which AI can do this is a long way off.
To find out more about the working world of UX design, try our free UX design short course.
And be sure to check out the top 15 ChatGPT prompts for UX design.
If you found this article interesting, then you’ll likely enjoy these: