What is the current role of AI in graphic design as of January 2025?

An introduction to any essay or article usually begins with the basics, providing the context or the outline of the article to come. This is a very good idea. Of course, the first thing an article should do is onboard its readers. But here is the thing; with the topic postulated in the H1 tag above, the basics are un-basic, the context is confused, and the outline is smudgy…

By now, everyone has read at least one article chastising or venerating our new AI overlords. Everyone is tired of hearing how it’s going to replace every profession, followed eventually by every person. Equally, everyone is tired of hearing the warning cries of the AI-fearing Luddite masses telling us to turn back now.

Everyone knows the basics, that context, these outlines.

But how might AI currently be affecting a topic closer to home for me, personally? How is it currently affecting the realm of graphic design?

Well, the thing is, I really don’t believe that it is right now…

Of course, the threat is there, threateningly threatening to morph into an even bigger threat. As far as us designers are concerned, AI is most definitely circling the edges of our walls. But it is hardly looking menacing as of January 2025. Achilles and his Myrmidons, these AI design applications are not…

The past two years, AI has taken a starring role on the public stage; it has been a rare moment that someone has tried to understate its influence. Sure, there have been lambasters a-plenty, but understaters? Nah!

But…

And it’s a big but…

In my opinion; AI tools have not impacted the world of graphic design in the slightest...

And this article is about why.

Consider yourself onboarded.

The weapons at hand

AI is usually delivered to us through apps. Some you will have heard of: ChatGPT, Siri, Alexa, DALL-E, Midjourney, Grammarly, etc. A lot of these are household names at this point. In the short time they have been in our lives, honestly, they have elevated them. Back to the Future II’s vision of what we had to look forward to was miles off; they got nowhere near predicting how useful something like ChatGPT could be. Imagine Google but he’s your friend, he knows everything, and he thinks you’re cool! That’s ChatGPT!

ChatGPT has made cheating on essays and sending 500 different job applications a breeze for university students and deadbeats respectively, everywhere. Thank you, ChatGPT, thank you.

As tools, these AI apps are incredibly useful. They have enhanced our lives. But is there a tool on the market right now as comparatively influential to graphic designers?

No.

There isn’t.

There is not.

Il n'y a pas.

There are no AI tools on the market right this second, that can help myself, Mench or even Paul (the designers here at Formation) make designs at the same degree of efficiency and efficacy that ChatGPT can write words. We have to do everything by our own hands, mouses and eyes. Jesus, is this the stone age?

My soon to be betrothed’s father is a developer. And he is constantly badgering on at me, the way badgers often do, to embrace AI in my work so I can ride the wave and not get left behind… A topic of conversation famously known to be very popular in the badger community…

Well…

I’m out here… In the water… Waiting… So where the hell is that wave I was promised???

Did that swarm of badgers lie to me?

A common assumption we graphic designers often hear is how AI image generation is somehow going to help us with our jobs or even outright replace us... The DALL-Es, the Midjourneys, the Adobe Fireflies and now the ChatGPTs. Well, we’re not illustrators and we’re not concept artists. So, being able to make nice pictures isn’t really that much use to us.

What we would really need is an app that can do layouts, be they social media graphics, brochures, or even whole website designs. We would come in afterwards as expert reviewers, mark the AI’s work and make any necessary changes.

However, there is no tool on the market that is capable of doing such things.

A few months ago, when I first looked into these so-called AI graphic design tools, to give myself a head start against this theorised new generation of AI-powered super-graphic designers, I found that the world of graphic design AI had not just missed the whistle, had not just failed to make the starting line, but had yet to be told that they were going to participate in a race…

To see what I mean, let’s take a look at some of the supposed AI graphic design tools currently available.

Weapon #1: Image Generators

We have touched on these already, we’re talking your DALL-Es, your Midjourneys, your Adobe Fireflies and now, your ChatGPTs, which has recently been blessed with an image making feature of its own.

I actually think that there is some confusion surrounding these types of apps and how they might be useful to a graphic designer. These image generation apps generate images, not graphics. And how useful are often-faulty, photo-unrealistic images to a graphic designer? Well, not very.

These apps have essentially killed off concept designers already. Anyone needing storyboards or character designs or copyright infringing fan-art of their favourite Pokémon can get that for £10 a month now. But as graphic designers, that’s just another £10 we can keep by not buying that Midjourney subscription.

We don’t need an image generator. Not only do image generators fail to provide us with the AI augmented layout designing feature we would actually need in our job roles, but even using an image generator for its intended purpose (image generation) is of little use to us.

If the client doesn’t provide us with images, there are stock photo sites where a plethora of suitable images is readily available to download. It’s a gamble whether apps like DALL-E or Midjourney will accurately generate what’s been prompted anyway, and even if they do, AI images are often quite unconvincingly AI-looking to cut any mustard whatsoever. To designers, image generators are a new, worse solution to a problem we've already solved.

I personally have used an AI image generator in my design once and that was only because I wasn’t logged into Shutterstock at the time, but did happen to be logged into my girlfriend’s father’s Midjourney… That old badger seems to be coming up a lot in this article…

It proved an illuminating experiment. But I could have got an employee stock photo from a million other places on the internet…

We don’t need an image generator. Really, we don’t. We need something that can do layouts, and an image generator is not that.

Weapon #2: Large Language Models (LLMs)

That’s things like ChatGPT, BERT and Claude.

Again, these are tools more useful in a different creative context, writing and copywriting, and even then, there are plenty of copywriters who would wince at the thought of using an LLM to do their job for them. (which is fair enough)

As a designer, admittedly, these tools do have their uses, if we need dummy copy or a list of random names drawn up or something equally place-holder-ey. ChatGPT is way faster than me at coming up with 20 different ways of saying John and Jane Doe. (altering superhero names slightly eg: Clark Kennet, or Wayne Bruce…)

Only once have I used ChatGPT to write copy and that was for some social media graphics that had yet to have their headings and sub-headings supplied by the client. We won’t go into how the client rejected these designs, citing the copy as the reason why – All ChatGPT’s fault, if you ask me.

Currently, if the graphic design AI tools are rocks, then ChatGPT and other LLMs are fighter jets… They are unmeasurably more advanced and more useful.

But alas, they are still of no significant use to designers.

ChatGPTs thoughts on the current influence of AI on graphic design

Weapon #3: Colormind

This is the first section where I'm going to single out one app in particular. Not because it is particularly bad, but because I think it quite usefully illustrates the current quality of so-called AI design apps.

Colormind is an app, an American app by the spelling of it, that allows designers to use AI to build colour palettes for them.

Now, I like Colormind; I think the idea is actually very good. It has the potential to be a very good tool. But in its current state, it has pretty much no use to anyone whatsoever.

This isn’t a review of an app, so I won’t go into too much detail as to how it works, but essentially, because the palettes are AI generated, the designer has no control over the colours being created by the generator. You can’t even pick a base colour to build the palette from. You click a button and 5 random colours come on screen… Great... Which just gets us back to where we were before Colormind existed, a situation in which it’s just better for us designers to use our own judgement when making a creative decision.

I see a lot of these types of tools on the market now. AI tools with no clear function or purpose, held aloft only by the buzz abbreviation itself: AI.

Colormind and apps like Colormind are not the AI tool we need to revolutionise the design profession.

Weapon #4: AI Logomakers

They’re bad, basically.

All of them.

Looka, Tailor Brands, Wix Logo Maker, Hatchful, LogoMakr, Brandmark, DesignEvo… None are worth the pixels they’re displayed on…

AI logomakers right now all function in a similar way. You make an account, you start the process, you input your business’s name, you pick your business sector, you pick some colours you like, some fonts you like, and maybe some icon styles you like, you click create…

THEN IT SMASHES THEM ALL TOGETHER!

It will then take you to a page full of variations of your logo, based on the suggestions you have input. Maybe then you can edit these logo options a bit, change a few colours around or something, change the basic composition and if you’re really nice and give the AI logo generator £9.99 a month, you can download your new, average logo as a PNG!

Now, here’s the thing: the process of creating a logo that I described above… How exactly, is this process more akin to an AI logo designing process and less akin to a browser-based Flash Player game? AI logomakers have all the function, customisability and aesthetic taste of a late 2000s “Dress up Barbie” game or “What logo are you?” interactive quiz.

I repeat my opening half-sentence; they are bad…

The comparison to the now-dead-and-buried multimedia software platform (Flash Player) may seem harsh and outdated. But honestly, it’s meant to be, because that’s exactly how AI logomakers feel.

I do not personally believe that these applications are run by AI as they do nothing that our computers and apps weren’t doing 10 years ago when they lacked AI functionality. What we are seeing is companies hoping to make a quick buck by jumping on the AI bandwagon for the sake of jumping on the AI bandwagon. This philosophy seems to be “If we say it’s AI, people will think it’s a new, exciting product”.

I have held my own private eulogy for concept artists everywhere. AI has essentially made them extinct in the span of two years. So why was AI able to replace something like concept art so easily and not logo design?

Logos require an incredible amount of conceptual articulation, combined with visual tightness. Or, logos have to be exact and AI struggles to be exact. There is room for manoeuvre in prompting concept art which has to get a vague idea across. However, there is no room for error in logo design. Logos must finalise and execute an idea, near-perfectly.

There are a million right answers to any logo design brief, but the answers must still all be right. It is the difference between having a machine churn out 1000s of gears and cogs vs having it do 0% tolerance machining. AI does not do perfect. It does estimates. It takes everything you have said, tries to understand it and then does its best to answer. But the end result is only ever an approximation, a guess. This process is not a good enough one to design logos.

I personally have a formula or system that allows me to design logos, so I am surprised that a similar system cannot be replicated with AI. This is one thought that I am honestly perturbed by. Once AI manages to make crisp, articulate vectors, capable of portraying reasoned concepts and do so via prompting, I think the design community might have some trouble on its hands. But for now, AI, leave the logo design to the experts, thank you.

Weapon #5: Canva

Now this one is interesting. People always expect me to hate Canva and see it as an easy answer or half-arsing but I really don’t. I like what it’s about and I think it’s a very effective tool.

In fact, despite not technically being an AI tool, Canva is probably the best weapon on the market that does what AI design tools should be trying to do.

For those of you who don’t know Canva, it provides about a kajillion graphic and animation templates that designers can use, customise, edit and download. These range from social graphics, to web-banners, to whole brochure templates.

Once selected, you can edit, cut, copy, crop, change colour, font choice, font size, images, pretty much everything. It’s not as powerful or as expansive as the Adobe products or Figma, but it’s not meant to be. It’s more of a tool for beginners. It allows the gap between complete novice and pro graphic designer to be bridged. And isn’t that what these AI tools should all do?

Yes. It is.

And I think in that respect, Canva is the best non-AI, AI tool on the market.

Weapon #6: Adobe’s AI features

I was not originally going to include this in the article. However, during my writing, Adobe held their annual Adobe MAX presentation. Cool name.

The event, as I’m sure you could have predicted, was very AI-heavy. Adobe’s products had several AI features before, but they were so comparatively limited in scope and use that it did not seem worth including them in this article. However, now that it looks like the most influential creative application suite in the world are about to dive headfirst into the open waters of AI, it might be useful for the last weapon featured here to focus on what is to come next in the world of AI and graphic design.

For the last few years, the Adobe AI features have been subtle and have varied in quality. The most useful Adobe AI features are the ones integrated into its most well-known app, Photoshop. One of the more advertised AI-powered features in Photoshop is the “Remove background” tool. Take a look at it in action here on a man with a hat!

With one click, you can remove the background of an image, which is why they named the feature “Remove background”.

I probably use this feature at least a few times a week. It is incredibly useful. The pre-Remove background way of cutting images out was to draw around them manually with the pen tool or to sketch around them with the brush tool. This could potentially take you 30 minutes, depending on how complex the image was.

Now it can be done with 1 click.

Another AI feature of note in Photoshop is its generative image function. This allows image editing via prompts.

Not bad and definitely not without its uses. This feature is very good for removing unwanted elements, correcting imperfections, expanding awkwardly cropped images, etc. But it has its limits.

These are both useful assistive features. Useful, admittedly, but not revolutionary. They are still a far cry from my coveted layout designing app I keep going on about as if I were some sort of furry black and white striped, four legged, long- snouted, forest dwelling creature… The name for which escapes me.

When will I get my wish? When will the AI layout designing function be there to change my life?!

…

Let’s take a look at what was mentioned at Adobe MAX, with a focus on one feature in particular: Project Remix a Lot, for Adobe Illustrator.

I was not expecting Project Remix a Lot, for Adobe Illustrator to do what it did in quite the scope that it did it. The first feature of Project Remix a Lot exhibited at Adobe MAX was the sketch to design feature. It works as it sounds. By taking a picture of your sketch, Adobe Illustrator can recreate a fully editable, vectorised artwork, based on it.

Cool. Quite impressive, actually...

But that’s not all! There is another feature to Project Remix a Lot…

It has a layout variations function…

While a base design is selected, you can instruct Adobe Illustrator to remake the design in different scales. This means that in seconds, you could potentially re-create your designs across any number of format settings. That’s a whole campaign based on one design. And the designs are all fully editable so any mistakes made by the AI’s guesswork can be amended by the designer!

We are not there yet, but this is a step in the right direction of where AI for graphic design needs to go. Layout generation.

I finally feel seen.

We are not there yet. We are not even at the foot of the golden hill. But it looks like Adobe is putting on its hiking gear and mountaineering equipment…

I wonder where they will go…

To Conclude:

So, how influential is AI in the world of modern graphic design today?

Well…

The question is interesting, especially after two years of the AI drug being on the market. The shrouded, sexy muse of AI has slowly disrobed, revealing not quite as attractive a figure as we were promised. Or, now that the whole “AI is going to take over everyone’s job” thing has died down, people are a lot less likely to predict the demise of every profession. Bus and taxi drivers are still apparently first on the list to go the way of the Dodo… However, mysteriously, a 7-year-old Willem still remembers people much taller and seemingly much wiser than himself at the time, making the exact same bold statement 20 years ago…

We are in a situation where it’s still best for graphic designers to rely on our own skills and understanding of graphic design, which, at this moment, are superior to AI’s skills and understanding of graphic design. We may be in a good tactical position on the top of the hill today, but at the speed AI is developing, we must still be prepared for that to eventually change.

When/if that happens, we can’t be like the Luddites smashing up machines, or the Church banning the printing of certain books. When new technology comes, we have a choice, we can adapt or die. But for today at least, not only do we not have to adapt, there is nothing we can adapt to.

Right now, AI tools are on the market for graphic design, and they range in quality from ok to useless.

When I first looked into AI graphic design tools to give myself a knife-like edge on my competition, I found only a dulled spoon…