On April 21, at the main Ural IT conference DUMP 2023, JetStyle designer Lesha Perminov spoke about his experience with neural networks and received a prize for the best report in the Design section. The topic is relevant not only for designers but also for the business they work for. Therefore, we have collected the main theses of Lesha’s speech and shared his presentation.
1. Neural networks are not new; the concept appeared in 1943, and the first algorithm was introduced in 1949. But in the last year, they became popular because they learned how to create art. Three main neural networks are especially cool: Midjourney, Stable Diffusion and DALL-E.
2. Neural networks are constantly evolving, and everything goes to the fact that there will be one cyber-powerful universal neural network that can do everything. But for now, many barriers remain: problems with anatomy, physics, interpretation of metaphors, and various random neuro mutations that will inevitably accompany you in almost every generation.
3. A neural network can create a beautiful interface, but this is only part of the designer’s job – it will not analyze the target audience or UX tests. She will find a thousand sites on the Internet and do something similar.
4. A neural network is not a magic “Make it beautiful” button. This is a designer’s tool, his assistant, which automates the routine and throws ideas. Therefore, artificial intelligence should be treated precisely as a tool that needs to be mastered and applied – both in combination with familiar programs and independently or in combination with a neural network + neural network.
5. The neural network does not guarantee speed. Some tasks can be solved 2–3 times faster
with it, for example, generating simple, generalized compositions or preparing a mood board with concept art. And some tasks are easier to do with more familiar tools – for example, drawing specific objects that require precise execution, or creating a series of visuals in the same style, or showing an object in several projections. When choosing a tool, you need to go from the task and not use the neural network with or without it to be on the hype.
6. To remain in demand as a designer, you must constantly develop, learn new things and get out of your comfort zone. Are you afraid that a neural network will replace you? No, you are more likely to be replaced by another designer who works with a neural network.