There is a famous anecdote about the Coca-Cola marketing campaign in Saudi Arabia, which brought unexpected results. The poster consisted of three parts: a man lying and resting in a desert in the first one. In the second picture, he drinks a bottle of cola, and in the third one, he keeps on moving through the desert. What could have gone wrong?
The problem was that people in Arab countries are not only reading from right to left, but they also perceive the visuals in the same way. Therefore, the audience got the complete opposite message. They saw a man who drank cola and needed to rest after that, while the beverage was conceived to be a refreshing drink by advertisers.
Improving UX with ChatGPT may have a similar result due to its limitations. Now let us look at the use cases where AI and UX design may not be the best combination.
One of the first as well as important steps in the work of a UX designer is user testing. The process involves gathering feedback from users on a product or service and analyzing it to improve the user experience and identify any issues that need improvement.
User testing requires real human participants to provide feedback on the product, and it goes without saying that ChatGPT cannot speak for users. Besides, ChatGPT is incapable of accurately testing certain aspects of UX, such as emotional responses or non-verbal cues.
Indeed, ChatGPT can mimic human response and behavior, but this AI is far from qualitatively simulating the complexity of real user behavior and experiences. This makes it less reliable for testing purposes, particularly when it comes to testing the usability and accessibility of a product.
But what about the situation when a designer has already gathered user feedback and needs to analyze it? Well, while ChatGPT can generate reports based on the data, it is not a substitute for human analysis and interpretation. A designer's ability to analyze and understand user feedback goes beyond just reading raw data; it requires empathy and an understanding of human behavior.
Creation of information architecture
It seems like organizing and structuring content is also one of the strengths of AI, and UX design process could benefit from it. However, information architecture requires a deep understanding of user behavior and information hierarchy, which is not something that can be effectively replicated by a machine learning model.
Still not convinced? All right, let’s look at the example of such a situation: