AI applications in UX/UI design

The site, the landing page, and the app are the points of contact between customers and the company, and, as a rule, the more convenient and "friendly" they are, the higher the user engagement and loyalty, and this translates to a high conversion rate. For example, Baymard Institute, a web research company, found that conversion rates at a large e-commerce site can be increased by an average of 35% if checkout pages have a better design. The UX/UI design is directly responsible for this.

For a long time the trend of minimal variation in interface design was maintained: thousands of sites and applications were created with an eye toward a kind of safety and sterility — with similar buttons, fonts, UX patterns. Later it became obvious that this approach does not create a strong connection with the self-identity of a particular brand, and today designers strive to create interfaces that reflect individuality. At the same time in the last few years in UX/UI there is a growing interest in solutions based on Artificial Intelligence (AI). Let's look at the interface design challenges where experts are increasingly using AI, and speculate about whether AI helps achieve a new level of individuality in the creation of digital products, and also whether it can compete with UX/UI designers.

AI application areas for the interface design

Machine Learning and Artificial Intelligence technologies are so deeply embedded in the programs designers use everyday that they often stop noticing their presence. There are different scenarios for using AI in interface design. Typical tasks include, for example, image resizing and automatic correction, when an AI-powered app (e.g., Remove) independently determines which part of the image should be cropped or modified. In some applications (Uizard, etc.), AI turns hand-drawn wireframes (low-accuracy interface layouts) for a mobile or web application into a ready-made prototype: the technology analyzes the elements, the location of figures or images, and creates a ready-made file based on the analysis. The intelligent features of such an app also allow you to "try on" the graphics and style you like for your project, using the screenshot as a reference. For other basic tasks, designers also often resort to site builders like Wix ADI, which automatically generates websites based on business information from public sources.

In addition, AI can be used to quickly improve the uniqueness of text through automatic rewriting, or convert audio to text, and prepare photo descriptions, which is especially useful for online stores and other websites. AI solutions are also employed when working with visual content. For example, neural networks are applied to draw the accompanying graphics for online articles and are able to generate more ideas per unit time (examples include the DALL-E 2 neural network). Deepfake technology makes a realistic substitution of an object in the video with the ability to automatically overlay masks and special effects (however, it should be noted that this technology has found the greatest use today to discredit individuals).

AI is also helpful for finding references. Interface designers often use references (supporting images) for inspiration (to discover new styles, visual design techniques, etc.), as well as to roughly visualize the result of their work and communicate their vision to their colleagues. In this regard, services (including the famous Pinterest) that build a user's feed based on their search query are incredibly handy — moreover, with each new query the system also updates the content.

Today UX/UI designers often use AI capabilities to generate on-demand content. Let's imagine that a website is supposed to have a lot of photos of people, for example, showing avatars of users of some service. There is a great risk of violating ethical norms, because it is strictly forbidden to use photos of real people without their permission, while it takes a lot of time to check each photo’s legitimate use. An alternative is to use an AI-based generator of fake faces that builds new portraits from parts of other images (for example StyleGAN generative adversarial network from Nvidia). And it is almost impossible to distinguish a fake photo: 90% of fake images are not recognized by an ordinary person, and 50% — by an experienced photographer. Sometimes AI makes mistakes, and images might contain bugs such as a wrongly curved contour, questionable hair or eye color, etc. However, programs learn quickly, and the probability of such inaccuracies is nearing zero.

As you can see, the capabilities of Artificial Intelligence do cope with a number of tasks in the design of interfaces, which greatly accelerates the project implementation, but the result still requires thorough checks and — in most cases — revision.

UX/UI designer vs. Artificial Intelligence

Since AI technology is so self-sufficient, it makes sense to ask, "what is it that AI can't compete with humans?" The answer, in fact, is simple. Artificial Intelligence can’t understand a person's idea if it hasn't encountered a similar case before: it only gathers information, analyzes it, and makes assumptions based on the patterns identified.

The question of whether AI has intelligence or not is quite philosophical. The presence of intelligence is determined only by what people themselves understand it to be. Even if we are talking about human intelligence, it is first necessary to define its attributes, since there is no single universally accepted list of exactly what makes us intelligent.

Until AI learns to understand people's motives, it will be difficult for it to compete with UX/UI designers. All the AI-powered programs evidently simplify and speed up the work, but the final decision is always up to the designer anyway. The above-mentioned trend to move away from standardized templates along with the desire to create a unique interface requires designers to deeply understand the intricacies of the customer's business.

* * *

AI solutions make the work of UX/UI designers much easier. Those routine tasks that take many hours to perform are completed in a matter of minutes. Meanwhile, when customers are trying to bring something recognizable and individual to their products’ interface, understanding their motivation becomes the main requirement for the designer. In this regard, all AI programs remain auxiliary tools used by designers to create interfaces. However, with the current active development of AI, it is very likely that in the near future such technologies will help shift from creating applications for generalized groups of customers into designing the personal user experience.