The Importance of Human Input in UX Research

A woman stands in front of a projector, with code illuminating her face
Source: Pexels

I’ll be honest — I’m not a fan of personas. In December, I published an article entitled Stop Using Personas, and Start Talking to People. I’ve worked in the industry long enough to know the road to hell is paved with good intentions. As such, I’m convinced that too often, personas are used in place of real research and are often based on dangerous assumptions. Hence my general dislike of them.

I recently read a very persuasive post on LinkedIn about how UX Designers can leverage ChatGPT — which they absolutely should, under the right circumstances.

My main issue was with one point, and one point only really: I don’t believe ChatCPT should ever be used to generate personas.

Here’s why.

Trust the (manual) process

Two people talking to one another in a well let room
Source: Pexels

AI has the potential to disrupt our workflows, for better or for worse, transforming our approaches to well-established working norms. This will undoubtedly lead to temptation. Why bother doing all the heavy lifting when AI can do it for us? The problem is, as of now, using AI to generate personas will come up short in a number of ways.

In recent months, there has been a growing trend towards using AI language models like ChatGPT to generate personas quickly and easily. But at what cost? While AI technology can offer many benefits in UX research, it is not the best approach for creating personas. In fact, using ChatGPT to generate personas can lead to inaccurate or unrealistic representations of users, which can ultimately harm the design process.

“It doesn’t work this way. You get requirements and build personas by talking to users and stakeholders,” Nick Groeneveld of The Designers Toolbox said.

Empathy and understanding the idiosyncrasies of human nature is the very reason UX exists as a discipline. It stands to reason that, for now at least, humans are better at understanding humans rather than a language model still in its infancy. Personas are only as good as the people who help create them, even then, they have the potential to be flawed and biased.

Personas are not just a collection of demographic data and personality traits; they are meant to capture the unique behaviours, motivations, and pain points of real users. Without direct input from users, personas are just assumptions and stereotypes that do not reflect the reality of user needs and expectations.

A dangerous precedent?

So not only are personas mostly assumption based anyway, potentially more so with the adoption of AI; but they’re potentially assumption based to an extreme we haven’t yet comprehended. Fundamentally they’re based on a machine language learning which is still in its infancy. That isn’t the basis for good, honest, transparent research.

Even the most die-hard fan of ChatGPT will admit it still lacks the necessary understanding of human behaviour, emotions, and context to generate accurate personas. While it can generate text that sounds like it was written by a human, it cannot fully comprehend the nuances of human communication or the complex interactions between users and technology.

For example, ChatGPT may not be able to distinguish between a frustrated user and a satisfied user who uses negative language, or between a user who values speed over accuracy and a user who values accuracy over speed.

Moreover, AI models are trained on large datasets, which can contain biases and stereotypes that are perpetuated through the AI’s output. For example, ChatGPT may generate personas that reinforce gender or racial stereotypes, or that overlook the needs of underrepresented user groups.

Uniquely Human

A collection of people walking across a pedestrian crossing
Source: Pexels

Using AI technology to generate personas may also create a false sense of efficiency and productivity, which can lead to shortcuts in the research process. UX research requires a significant investment of time and resources to conduct effective user research and data analysis. Using ChatGPT to generate personas may seem like a quick and easy way to gather insights, but it cannot replace the value of direct communication with real users.

In contrast, a human-centered approach to persona creation involves engaging with users directly through interviews, surveys, and usability testing. By gathering insights from real users, designers and developers can create more realistic and accurate personas that reflect the needs and goals of their target audience. This approach also allows for a deeper understanding of user behaviours and preferences, which can lead to better design decisions and more user-friendly products.

A human-centered approach to persona creation is the fundamental basis for good design. I still don’t like how personas are used, but I especially don’t like the prospect of AI generating them in their droves. Talking to users, understanding their needs and learning more about their lives makes me feel incredibly privileged. It’s what I spend a lot of my research time doing, and it’s what UX designers need to continue to do.

So forgive me if I don’t get caught up in the hype around AI-generated user personas. I think I’ll take just a little longer, and let the people speak.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *