Why AI for UX is worse than nothing

Why ChatGPT for personas will leave your product team worse off than if they had none at all

Joanna Weber
Bootcamp
3 min readApr 24, 2023

--

“This (persona) is actually worse than nothing”, Jason Godesky says of an AI-generated persona, and here’s why.

Sarah Thompson, age 35
Occupation: Sales Manager
Wants to find a hotel close to a conference venue

is garbage. It doesn’t tell us anything about WHY Sarah wants to find a hotel close to a conference venue, and that is crucial to every decision you make from how to build the app to where to place the advert.

If Sarah wants to find a hotel close to the conference venue because she’s a social animal going to lots of late night afterparties, she’ll have very different needs to a tired introvert with mobility issues.

The persona types I use most frequently are a proto-persona, and a qualitative empathy map persona.

An example of a qualitative empathy map persona — template from alexandercowan.com

A proto-persona is a list of assumptions you/the team has about the person using (or buying) your product.

That’s why you can’t outsource it. You have to do it yourself in order to think about and write down what assumptions you have, and then identify which are your riskiest assumptions.

What do you take for granted that you know about your customers but, when you look at the list honestly, you have no real data to back up that assumption?

That gives you a list of the things you need to test through interviews. That’s why we do it — not to have a pretty stock photo and a nice-looking poster on the wall.

It’s a practical exercise, ideally conducted with product managers, engineers and the whole team, to discover your unknown-unknowns.

Then you test it and compare your interview transcripts with your proto-persona. What did get wrong? If you map your summarised transcripts to a persona template and you find gaps in your observations about the user’s cognitive process, behaviour, wants and fears, (or “thinks/sees/feels/does”) then you may need to ask more questions.

That’s what it’s for: to show you what you don’t know, and then check that you answered the key questions.

“ChatGPT can’t do UX research for you. That’s fundamentally outside of what it can do. It’s beyond what any AI can do. If you take the users out of it, what you’re doing isn’t user experience design.”

Godesky does suggest ways that AI can help UXers and product teams, such as showing how other people have tackled a problem or generating a list of competitors — but each of these functions show up ChatGPT’s true nature: it’s a natural language search engine, capable of the basic functions of desk research. It is not a magic box that replaces the essential work of humans.

Sorry, Skynet, we live to work another day — but if you have any inspirational prompts for the next UX article, I’m all ears.

--

--

Bootcamp
Bootcamp

Published in Bootcamp

From idea to product, one lesson at a time. Bootcamp is a collection of resources and opinion pieces about UX, UI, and Product. To submit your story: https://tinyurl.com/bootspub1

Joanna Weber
Joanna Weber

Written by Joanna Weber

UX research and product development | author of Last Mile