We need to talk about AI because it’s a game-changer
AI will not replace you as a designer, but a designer who’s using AI will replace you.
There’s so much talk about Chat GPT, Dall-e, and other AI-assisted tools. So I was curious and tested ChatGPT for a week to see if it was really that good.
In this article, I will show you how we designers can use the latest hottest AI tool, ChatGPT.
Let’s ask chatGPT itself.
It seems I wasn’t specific enough with my question. So let’s ask ChatGPT again, now focusing on UX designers.
According to ChatGPT, it can help assist user experience (UX) designers with the followings:
For example, a UX designer could input a prompt into ChatGPT, such as
“Design the onboarding flow for a new task management app” and ChatGPT could generate a list of ideas and suggestions for the onboarding flow, such as introducing the app’s core features and functionality, highlighting the benefits of using the app, and providing tips and best practices for using the app effectively.
Prompt:
Write UX copy for a new task management app’s landing page. Highlight the benefits of using the app, such as increased productivity and organization
Output:
Then I asked ChatGPT to make it shorter. Here’s what it said:
This is much better, simpler, and snappier.
You can ask ChatGPT to keep refining it until you’re happy with the result.
Prompt:
Design the onboarding flow for a new task management app
Output:
Guess what? ChatGPT is a conversational AI.
So unlike a Google search, you don’t need to enter the prompt again, you just follow up on the previous prompt, and ChatGPT just magically knows what you meant.
I asked ChatGPT to add a few more items to the list, and it did. See it below.
Now, is this perfect?
No. It’s not perfect.
However, it makes you think. In a few seconds, you will have 10+ items regarding onboarding for a new task management app.
But can it create wireframes?
Let’s find out!
Prompt:
Create a 4 screens wireframe on bulletpoint 13
Output:
Well, this was hit-and-miss, for sure.
ChatGPT didn’t draw wireframes. And the output is super flawed.
You don’t need to be a seasoned designer to spot mistakes easily.
Prompt:
Generate user flow ideas for an e-commerce app
Output:
This is a pretty good starting flow if you don’t know how to start a project.
If you’ve done an e-commerce project before, you’d be able to write a list very similar (and add more items to it) easily, though.
Prompt:
Give me the 5 most popular e-commerce websites with links
Output:
I wanted to see if ChatGPT can do some sort of competitor analysis. So, I asked.
Prompt:
Do a competitor analysis of Amazon’s 5 biggest competitor
Output:
You could have arrived at the same result by doing a simple Google search, but this is a good starting point, and you saved a few minutes.
Each point needs further dive in.
I wanted to test if ChatGPT can help me write a research plan.
Prompt:
Write me an introduction to usability test, briefing participants what’s going to happen during the test. Imitate Steve Krug.
Output:
I need to say this is a good starting point that you’d be able to copy-paste and make it your own by adding all the little details you want to include.
I’m not 100% sure how much ChatGPT could imitate Steve Krug. But you can try to test it with other, more well-known figures, such as politicians.
Prompt:
Write me a usability testing script for a pet sitter iOS application. I want to find out what makes users use a pet sitter
Output:
The output is different from what I had in mind. This output looks more like a usability testing task list that you can ask from users.
It seems my prompt wasn’t clear enough. So let’s try it again:
Prompt:
I’d like to interview pet owners. I’d like to find out what makes them use a pet sitter. Write me 15 interview questions I can ask pet owners.
Output:
The output is much closer to what I imagined initially.
The questions are okay.
However, I was expecting ChatGPT to display more questions answering the prompt. I would’ve liked to see more questions similar to #2. Questions that could really dive deep into pet owners’ motivation to search for a pet sitter.
Prompt:
Write me a research plan for a UX discovery research of the pet sitter market.
Output:
Again, the output is a good start.
For better results, I’d recommend predetermining the research goals and objectives and the target audience.
Prompt:
Summarize the following text in 3 paragraphs: [paste interview transcript]
Output:
I inserted one of my articles to see if ChatGPT can handle it.
I must say, the summary is accurate. Well done, AI!
Prompt:
Create a Google Design sprint workshop outline for 5 days
Output:
As ChatGPT says, the output is a generic outline.
From here, you can ask ChatGPT to elaborate on each day by adding more meat to them.
As someone, who’d run Design sprints multiple times in the past, I’d be better off writing the details myself instead of asking ChatGPT. However, this is a great outline to get started if you’re entirely new to it.
To successfully run a design sprint, you do need to do some background reading on the topic, though.
Prompt:
Design a logo for a beauty therapist
Output:
ChatGPT can’t design a logo for you.
However, it describes what a specific logo should look like. It even offers font suggestions to complete the overall brand identity.
Goodbye Fiverr!
Ok, not yet. But maybe, soon?
Prompt:
Write a 3 paragraphs cover letter for this position: [inster link to job ad]
I used a Lead UX designer job ad advertised on Indeed.
Output:
No wonder my previous video tanked on YouTube (imagine cricket sound).
There’s no excuse for not writing cover letters anymore.
It takes a minute; you massage it for another 5 minutes so that it’s actually true, and you’re done.
Finally, I tested if ChatGPT can suggest design tasks to practice.
Prompt:
Give me 10 UX design prompts so that I can practice design challenges and add them to my design portfolio
Output:
ChatGPT nailed this task.
As ChatGPT doesn’t make everyone a great writer overnight, Dall-e doesn’t make everyone a great artist overnight.
Even though there are Grammarly, Wordtune, Quillbot, and many other AI writing tools nowadays, they still haven’t replaced human editors.
It’s similar to Canva or Adobe Express; these tools haven’t replaced graphic designers either.
Sure, some business owners would rather use cheaper labour, but that’s ok.
There’ll always be a market for human-made products. It’s kind of like a hand-made item. It’s a premium.
Would you pay a premium for a handmade knitted cardigan?
I know, I would.
I predict, soon enough, AI will be able to draw simple user flows and design screens.
You will enter a prompt, such as “design a login flow happy path for a pet sitter iOS application”. Then ChatGPT or Dall-e will just do exactly that.
ChatGPT is something I never knew I needed.
But let’s always remember its constraints when we use ChatGPT to assist our work.
It even tells us about its limitations:
“It’s important to note that ChatGPT is a tool, and its output should be used as a starting point for further refinement and editing, rather than being used directly in a final design.” — ChatGPT
Never take its output at face value because ChatGPT is still learning.
It still produces incorrect answers, especially when you replace your Google search with ChatGPT prompts.
But instead, take its output as a starting point.
If you haven’t tried it yet, go ahead.
Let me know if you find another use case for it. I’d love to hear about your discovery.
This article is the written version of my YouTube video: