The Algorithm in the Room: When Tools Become Distractions

There is a particular kind of weary sigh I only hear from creative professionals being told by a manager to “just use AI™ for it”, or being gently “encouraged” to bring AI into their workflow: because TechBros™. It is a sigh born of the realisation that the person in charge has mistaken a firehose for a fountain pen.

Lately, the creative industry has felt less like a studio and more like a laboratory where the scientists are obsessed with speed over substance. There is a frantic rush to ram Large Language Models (LLMs) into every corner of the workflow. However, this forced integration is doing something rather counterproductive: it is slowing down creative thought by cluttering the workspace with digital noise.

The Great Definition Muddle

The frustration often stems from a lack of clarity. We are told we must embrace “AI”, but no one seems able to define where the “tool” ends and the “intelligence” begins.

If I use 'Select > Same > Fill Colour' in Adobe Illustrator to update forty elements at once, am I using AI? Technically, it is a programmatic algorithm performing a task based on specific parameters. When I export a press-ready PDF from InDesign using European Press Standards, I am relying on an incredibly clever set of instructions to ensure the physical world matches the digital one.

We have been using “clever algorithms” to assist our design work for decades. We simply called them “features” or “shortcuts”. They were invisible, reliable, and — most importantly — they stayed out of the way. They didn’t hallucinate a fifth leg on a stock photo or try to write “edgy” copy that sounds like a corporate brochure having a mid-life crisis.

The Curator of “Good”

The fundamental skill of any seasoned creative isn’t just the ability to push pixels or string sentences together; it is the ability to know what “good” looks like. We are paid for our taste, our discernment, and our ability to filter out the mediocre until only the solution remains.

In the current climate, we are increasingly being recast as “curators” of machine-generated output. But here is the rub: the more “AI” we ram into the mix, the harder that curation becomes.

Instead of starting with a blank canvas and a clear intent, we are presented with a relentless firehose of “content”. We are tasked with finding the needle in the haystack, except the haystack is growing at an exponential rate and most of the needles are actually just weird-looking paperclips with floating eyes asking “ do you need a hand with that?”. Adding more AI to the process doesn’t make us faster; it just makes the search for quality more laborious. We aren’t being given better tools; we’re being given more hay to move.

Tool vs. Distraction

The differentiation isn’t necessarily in the code, but in the intent.

Old-school programmatic algorithms are tools. They are predictable. They take a tedious, repeatable task and automate it so the human brain can focus on the “why”. They facilitate the flow state — something particularly vital for the neurodivergent creative who may already be battling a cacophony of sensory input.

LLMs, in their current “rammed-in” state, are often distractions. They require “prompt engineering” (a fancy term for begging a machine to be less mediocre) and constant fact-checking. Instead of removing a hurdle, they add a layer of management. We are no longer just designing; we are supervising a digital intern that has read the entire internet but understood none of it.

The Cost of Forced Efficiency

When leadership forces these tools into the workflow, they aren’t just asking for speed; they are disrupting the delicate internal process of creative problem-solving. For many of us, the “slow” part of creativity — the thinking, the sketching, the staring blankly at a wall while our brain connects two disparate ideas — is actually the most efficient part of the process.

Perhaps it’s time we stopped asking how much AI we can fit into the workflow, and started asking whether the “solution” is actually just making the problem more complicated. Personally, I’d quite like to go back to my “old-school” algorithms. At least they knew their place, and they never tried to “collaborate” with me on a Tuesday morning when I was just trying to get things done.