AI image generators: the end of graphic designers?
As technology continues to evolve, so does the way we design.
Our team of graphic designers tried AI image generators to measure the impact they could have on their work. So far, so good.
AI vs traditional graphic design methods
Let’s first consider the differences between AI design and traditional graphic design. Conventional design methods rely on designers’ creativity and experience to develop ideas and create visuals. Unfortunately, this method can be time-consuming, as it requires a lot of trial and error to find the right look and feel for the product.
On the other hand, AI design uses algorithms and machine learning to generate unique visuals. This is done by providing the AI with parameters, such as subject matter, colours, styles, shapes, and textures, which the AI then uses to generate images. This process is much faster than traditional design methods, particularly when using open-source software currently available for developing new imagery and compositing or montaging images.
Our experience using DALL-E
At Cobo, we have been trialling DALL-E, which comes from OpenAI’s pioneering GPT language models that can generate text. DALL-E – the name is a combination of Salvador Dalí and the Pixar character Wall-E – was trained on 400m pairs of images with text captions “scraped” from the internet. It essentially swaps images for text when given an instruction. Read more about DALL-E on the Openai blog.
When we command DALL-E to create a particular image, as by a human hand, the results are slightly different each time. It is in instant image creation that this AI has an immediate impact on our daily work. Visualising new graphic design ideas can be accelerated. Consequently, initiatives like Dall-E are exciting tools for extending our practice. It is like having an extremely efficient team member contribute to ideas early in a project.
Similar to pets doing ‘crazy’ or ‘cute’ things on Instagram and Facebook, DALL-E is awash with kooky combos of images: a still of Homer Simpson in the Blair Witch Project, a velociraptor working at a hotdog stand, a hamburger in the shape of a Rubik’s cube, the list goes on. We bet Douglas Adams would have loved it. But, for now, AI image generators are not fit for rolling out or executing a design concept in its finer details. You may want to put AI in charge of your social media images but not your brand guidelines.
OpenAI (2022) Dall·E: Creating images from text, OpenAI. OpenAI. Available at: https://openai.com/blog/dall-e/ (Accessed: February 17, 2023).