How can designers make use of AI?
Of course, we at Increo have had fun with the latest tools in the AI playbox. But it's serious in all the fun. Is there reason to be concerned? Not at first, thinks our UX designer Kristina Ermolenko.
There's a lot to be said about language models like ChatGPT, and the discussion is hot among those who work on producing — whether it's code or text. But what about designers? Are Midjourney and Stable Diffusion the hook on the door for graphic artists?
These are some of the AI tools
Dall-E — Generate images from text. Built on the GPT-3 model by OpenAI. Can generate a wide range of images and graphics, in a variety of different styles.
Midjourney — Text to image generator.
Stable Diffusion — Text to image generator.
ChatGPT — A chatbot built on OpenAI's GPT-3 model. Allows users to ask questions and further communicate with the AI in conversational form. Generate text in various forms, and also, for example, code.
Co-pilot — AI powered code assistant developed by OpenAI and GitHub. Copilot actively makes suggestions during writing code, does autofill, and can generate longer snippets based on comments. The AI is trained on open source from GitHub and also uses your context to provide tailored suggestions.
“It's definitely easy to think that the answer is yes when you see what kind of artwork these guys are capable of creating. I understand a lot of people are worried, but I'm an optimist. There are several reasons for that, says Kristina.
No strategist yet
One thing is that these tools need a creative force behind them. The AI must have an input it can generate a “response” to. In order to provide the right input, one must have an understanding of both the context in which one works, what the intended result should look like, and not least how the tool can be used.
Although these AIs are relatively easy to get something out of, it takes a bit to get something that matches what one wanted. It's a bit like Googling — getting the right hits is an art that must be learned with experience.
“The work we do is far more fundamental than just making something nice. There is a lot of insight behind making a design reinforce a business strategy. It takes research and knowledge of people and psychology, and I think it's a while until AI becomes as good at advising as it is at drawing.
Kristina, UX designer in Increo
The interesting question is thus more practical than philosophical: So, how can designers, with technology as it stands today, make use of AI in their work?
3 Useful uses for the designer
The possibilities are, of course, very many, but in Increo's small AI think tank we have landed on three obvious uses that are both low-threshold and valuable.
1. Exploring ideas
Even the most creative people need inspiration. Since we live by just being creative, it is easy to go “empty”. Then we need a sandbox to play in, options to choose from. With AI, you can more quickly take loose ideas “all the way out” to see if they're worth pursuing -- or discarding.
“These tools simply provide opportunities for a much higher rate of experimentation. We can try out both several and some more crazy ideas before we choose one to commit to, Kristina explains, adding:
“For me who love colors, AI is also a useful source of inspiration when it comes to color use. It's not a substitute for creativity, but more of a sidekick that keeps your own creativity going.
2. Next-level mockups
Everyone has seen them: boring mockups of everything from business cards to iPhones and coffee cups. Therefore, in fact, this does not need much explanation.
Don't underestimate a good mockup. Often, after all, they are an important part of the first impression. So why not make them a little more exciting? With Dall·E 2 and Stable Diffusion, you can stage photorealistic mockups that take your entire presentation to the next level.
3. Short down the road from wireframes to finished design
Kristina also believes that her UX design colleagues sit safely in the chair for a while longer. But she predicts they'll have it easier. Before long, she believes, we'll have artificial intelligence that could act as “autocomplete” for interaction design. Kinda like GitHub Copilot, but for UX.
— There are already services that convert hand-drawn wireframes into Sketch and React code, for example. As we stand on the cusp of an AI revolution, I think there will soon come a day when I'll just write “give me the best form” and then I'll be served what the AI knows is best practice.
And I don't think that day is far away.
