Experimenting with Looka’s AI tool to make a logo

Long read

So we tasked one of our designers to do a little experiment to try out the latest AI tool Looka for some studio fun. See her thoughts on the process and findings below:





I’ve been set this challenge and I must admit, my inner cynic (let’s call her Simone) has reared her ugly head. The AI tool company’s proposition is: “We’re here to make great design accessible and delightful for everyone… Everyone can make their business, side hustle or passion project look beautiful, even if they’re not a designer.”

The proposition is valiant enough. It’s the “delightful” and “look beautiful” that I’m personally sceptical about. I wonder if Looka created their own brand using their own AI tool – and if not, why not? But like an investigative journalist, I try to keep an open mind.

For context, the brief I’m applying this experiment to is my own side hustle that I run with a friend of mine. A jigsaw puzzle company called Rejig. I spent many a day stressing over typefaces, colours, logo options and puzzle collaborators. So if there was a tool that could have reduced some of those ‘out-of-hours’ hours, I’d have taken it, gladly.

First things first, Looka asks you for a brand name. Since naming is one of the hardest parts of any brand journey (you only need to watch The Apprentice contestants choose their team names to understand why) I feel I’m being a little short-changed here… Maybe a chat GPT plug-in can help with that. A quick brief to GPT and any optimism is dashed. Simone cracks a wry smile.





The next step is to choose which market category you belong in. I want to tell the AI that REJIG is an intersection between games and homeware. It’s a game that would feel at home on a coffee table. But I’m only allowed one category, so ‘Games’ it is.

You pick up to 3 colours you like and also have the option to choose a few symbols. Since all the “symbols” looked like 1990s clip art (and not in a good, way), I decided to skip this stage.





Next come logos, thick and fast. Streams of them, it’s endless. Despite having skipped the ‘symbols’ section, the logo options all feature one, all of which are about as relevant as a floppy disc.





Like all good AI tools, a few well-known marques are ripped off wholesale. Cynical Simone wonders whether REJIG should consider a merger with a Telecoms or tech company, just to make the logo selection-process stop.

But to be fair to the AI tool, taken in solitary, there are a few nice, grotesk typefaces in the mix. Credit where it’s due. I land on a wordmark.





There is an ungainly gap between the bar of the E and the J that I desperately want to rectify. Alas – that is not the name of this game. A handy ‘More Ideas’ tool kicks into gear (the AI equivalent of your boss taking control of the mouse and making your design explorations infinitely better), and options which are ever-so-slightly more up my street are produced…





I end up picking a logo which isn’t a million miles away from the own wordmark I created myself. The colours are even very closely matched to one of the test puzzle packaging options I made, (but ultimately didn’t go for).






So what of the experiment? I ended up with something similar to my own end logo, can I really be a sceptic?



My logo





The AI’s logo



In truth, I still am. I’m sceptical because design is hard, which means low-quality, visually cliched artwork exists in abundance. And that low-hanging fruit is the diet these AI’s are consuming and then re-producing back into the atmosphere.

I’m sceptical for the same reason why I’m sceptical about AI art and stock imagery… where when you type in ‘woman’ the images produced always appear to default to mostly model-like women running a dainty hand through their hair, or worse yet, scantily-clad adolescents reflecting society’s male-gaze.



I think if you as the end-user don’t have access to a design resource, these tools could be useful to you. But if that’s the case let’s hope you’ve got a good eye, so you can pick something tasteful out of the abundance of not-so-great options. In all other circumstances, until the technology is more considered, more discerning, and more grounded in design theory, it’s a hard no from me.