UX & AI on the Same Team
During 2023, AI became a trending topic and mass media was practically all over it. Individuals who were unfamiliar with this technology had the opportunity to use artificial intelligence tools, thanks to the simplification of user interfaces and the upheaval they caused. Social media platforms were flooded with tips on how to leverage these tools to increase productivity. There was undoubtedly a media hype. However, beyond all of this, something is true: there is a technology (or a technology set) that has the potential to fundamentally transform our way of working. This transformation will not only affect how society produces goods, products, and services but will also have an impact on UX design. In this article, I specifically want to refer to the impact AI has on the user experience design field.
Within the realm of user experience design, we can identify two main areas on which artificial intelligence is having and will continue to have a huge impact:
- The way in which we design: this is translated into an improvement that includes greater speed, the automation of tasks that do not add significant value (and that are often somewhat tedious), thus allowing design teams to have the space to delve into and improve the results of their creative tasks.
- The experience itself: there is another improvement that implies that people will have a more satisfying experience when interacting with our products or services.
Hereunder, I will focus on the first one of these areas, thus leaving the second one for another article.
Improvement in the Way We Design
Since ChatGPT has gained popularity, numerous publications have emerged on how to leverage various artificial intelligence tools to streamline Designers’ work. If you wish to delve deeper into this topic, Gustavo Soto Miño devotes an entire chapter to it in his book Diseño Artificial [Artificial Design], in which he reviews the different techniques used in experience design and how to enhance them through artificial intelligence.
I do not want to be reiterative with what has already been extensively discussed, but I do want to emphasize how artificial intelligence can improve our work in each stage of the experience design process. While there are many models and charts that represent the design process, they generally agree upon the existence of the discovery, ideation, prototyping, and evaluation phases. Although these phases occur concurrently and cyclically, we often represent them linearly for the sake of mental clarity.
Discovery
To discover means to find something that is unknown. In the case of design, it involves acquiring knowledge and then translating that knowledge into different artifacts that will guide us when making design decisions.
The heart of the discovery stage lies in obtaining information. Here, cautiousness is crucial. Artificial intelligence can provide us with a vast amount of information and even format it into artifacts or deliverables we commonly use, such as personas (archetypes), Jobs to be Done, storyboard scenes, insight reports, etc. However, the problem with such information is its quality, which is not bad necessarily, but it does come with the following problematic aspects:
- We cannot determine, at least today, the information source that feeds artificial intelligence.
- We cannot know whether it is a hallucination of artificial intelligence (i.e., events that did not occur or made-up data).
- Artificial intelligence tends to replicate the systematic mistakes of the ones who have trained it.
For this reason, at a personal level, I prefer not to use artificial intelligence as a reliable knowledge source. However, these tools can significantly help us during the discovery stage. They can assist us in drafting interview scripts, provide guidance for trend analysis, list relevant competitors for benchmarking, and provide us with a research framework. But the greatest value that artificial intelligence adds to this phase is during the final steps of a research. At that moment, we are faced with a large amount of data and information we need to give meaning to so as to turn it into knowledge. All this information, ranging from interview transcriptions, post-it notes on a board, free-text responses in surveys, to publications in specialized media and more, is unstructured, and AI precisely excels in understanding vast amounts of unstructured information, identifying patterns, and uncovering emergent trends. Therefore, when transforming information into knowledge, artificial intelligence becomes a valuable ally to reduce time and increase efficiency.
Ideation
Anyone who has tried ChatGPT has been able to experience the creative power of artificial intelligence. From asking it to make a recipe with the four ingredients in our refrigerator to asking for a story for children with our child as the superhero protagonist, the creativity of these systems just seems overwhelming. This creativity can be harnessed for ideation within the design process. As Nielsen mentioned in one of his articles, there is increasing evidence that shows that the combination of artificial intelligence and human intelligence is much more creative than each of them on their own.
Regarding the ideation workflow, there are several possible scenarios:
- AI creates, humans refine: AI can craft new and different ideas, while humans can contribute to refine them and make them more feasible with their creativity and experience.
- Humans develop drafts, AI refines: People can write drafts, and AI tools can be used to get feedback and improvement suggestions.
- Humans, AI, humans: Personally, and subjectively, this model has yielded the best results for me. It involves iterative cycles in which people develop ideas, receive feedback to improve them, iterate, and seek feedback again, in a cycle that concludes when the person determines that the best possible idea has been reached.
Regardless of the workflow model, the co-creation between AI and humans can be a powerful combination.
Prototyping
It is entirely feasible to provide artificial intelligence with our design system, including all of its components and usage rules, and to teach it the style of our illustrations, and the texts’ shape and tone. In this way, we can give the low-fidelity design of some interfaces to AI and allow it to develop the high-fidelity versions. Today, tools that can transform hand-drawn wireframes into high-fidelity screens already exist. There are also tools with which one can describe what is needed, and artificial intelligence will create one or multiple screens. Although these tools currently operate in a rudimentary manner, the rapid evolution of AI systems suggests that they will become much more powerful and precise in the near future.
I am not implying that the task of prototyping will come to an end, but the time and effort devoted to it will be significantly reduced.
Within prototyping, a new variant that holds great potential opens up: prototyping “with code.” There are several tools that can transform our designs into code and that are available for use. The line between design prototypes and a functioning software product is becoming less rigid.
I propose an experiment, write the following prompt in ChatGPT or Google Bard:
“Develop the HTML, CSS, and JavaScript for an onboarding process of a banking platform. Ask me anything you need to know to build it.”
The result will obviously be rudimentary, but it can be useful to test or refute a business hypothesis.
ChatGPT and Bard are text-based tools trained to handle a wide variety of topics. If artificial intelligence is specifically trained to design interfaces, with the same prompt, it will be able to craft all the screen flow templates with the components of our design system, our text style, and our illustrations. Afterwards, it will be possible to edit them with our preferred design program and to transform them into functional code.
Evaluation
If we follow Nielsen’s heuristics, and embrace a set of good practices, design patterns can be used to train AI. Consequently, artificial intelligence will be able to analyze our designs and identify potential issues.
Some companies even promise that they will train artificial intelligence to perform usability tests automatically. While I do not believe that it will replace real user testing in the short term, it will be possible to have an initial approximation to a system’s usability issues through tests conducted by “virtual users.”
AI can also streamline and improve A/B testing. In fact, it can run multiple tests simultaneously by automating the creation of variants and by measuring their performance in real time. Additionally, it can learn which patterns yield the best performance, making A/B testing increasingly efficient.
This allows design teams to make faster, data-driven decisions.
Will We Lose our Jobs?
Production is undergoing a transformation, and jobs will be affected. Some talk about redefining what work means or how to distribute the profits generated by artificial intelligence. However, I do not want to delve into that debate.
Designing digital products is much more than drawing screens. It involves understanding the business, identifying opportunities, and envisioning scenarios that meet people’s needs. We need to go down that road, create non-automatable added value, and become indispensable in the productive framework. Artificial intelligence is reshaping the design field. Many tasks traditionally performed by design teams will be delegated to artificial intelligence, particularly routine tasks that add little value. As Designers, we must change the way we do things, we cannot remain where we are, or technological advancements will render us obsolete.
Even for such tasks that will be delegated to artificial intelligence, human oversight will still be necessary. Therefore, ethics becomes crucial in design, since design teams will play a fundamental role in making responsible decisions about our products and their societal impact.
While it may sound like an optimistic vision of the future, artificial intelligence, if we learn to use it as an ally, has the potential to enrich and improve the work of design teams. It provides opportunities for creativity and continuous innovation in the digital design industry.
One thing is certain: in two years, companies will be seeking Designers with at least three years of experience co-designing with artificial intelligence. So, it would be a good idea to start using them now to stay ahead of other Designers and to stay current with the necessary skills within the design job market.
