(Artificial) life imitates art: AI and design

General-Purpose vs. Purpose-Built Tools

Let’s be clear: there are a lot of design-related AI tools out there, and not all of them offer an advantage over doing things yourself. So how do you choose the right tool? We believe that the main question you should be asking yourself is this:

“Do I want a jack-of-all-trades tool that I can use as inspiration for all kinds of projects, or do I want a virtual design assistant that is really good at doing one specific thing (completely)?”

A fitting example of this is website design. You could use general-purpose image generation tools like Midjourney and DALL-E to create all kinds of images, ranging from inspiring visuals to website design mock-ups. They are great at producing standalone basic images, but they won't be able to generate entire designs, and they cannot produce (legible) text.

On the other hand, purpose-built tools are great at creating (almost) ready-to-go results for specific use cases. However, they cannot help you beyond that: a website design tool won't create storyboards for an animated video. It's also not always clear whether these tools use AI or templates. For example, no-code platforms like Dorik or Bubble let you build websites or even full-stack apps without a single line of code, and it's not yet clear how their AI offerings will differ.

That being said, here are some up and coming tools for designers that definitely use generative AI:

• Magician for Figma generates icons, images and copy directly in Figma.
• Visual Electric is an image generator built with creatives in mind.
• Leonardo AI generates ‘production-quality’ visual assets for various disciplines.
• Mokker focuses on product photography and replacing backgrounds.

No matter which tool(s) you choose, you’ll still have to rely on your own expertise. Don’t trust artificial intelligence to do everything or know the context. Advanced as they may be, they cannot account for aspects like the responsiveness of a website or a customer’s brand guide yet.

Three Categories, Three Best-in-class Tools

Once you have decided whether you need a specific or a more general tool, there are three other criteria to consider: the quality of the result, the user experience, and the ‘feedbackability’, or the degree to which a tool will be able to tweak the results based on your comments. Let’s take a look at these elements and find out which tool is the best-in-class solution for each.

First up is the quality of the result – and what you are willing to sacrifice for it. Do you want a state-of-the-art tool that delivers the best possible output, or will something a tad less do as well? At the moment, Midjourney is the undisputed king when it comes to image quality. However, it is not without its drawbacks. It works entirely through Discord, a platform for instant messaging and (video) calling that mostly caters to gamers, and it does not handle feedback that well.

As a professional designer, Midjourney will likely feel far too limiting when it comes to the user experience. You will have to interrupt your workflow and may have to start over entirely when prompts are misunderstood. If you want to stay in familiar territory, we highly recommend giving Adobe Photoshop’s AI features a try. Tools like generative fill and generative expand give you fine-grained control over the output, and you won’t have to learn how to work with new applications.

For those who want to be able to tweak the output but prefer a conversational style, the ‘feedbackability’ we mentioned earlier may be a deciding factor. In this case, we recommend using DALL-E as part of ChatGPT Plus. This requires a subscription, but it’s a true game-changer. Unlike other tools, you’ll be able to tweak its output based on your feedback, both through text and by giving it other images that match your desired output more closely.