AI: no substitute for human thought or design

Rob and Gaby consider the differences between AI and human problem solving.

A rusty robot connected to a globe.

AI needs rules to function. However, to be truly inventive, we must travel to places where the rules don’t apply.

It’s an attractive idea to run a prompt through a machine that promises to fix your problems. But are the solutions it generates appropriate? Are they well-considered? Can they really be implemented by a business or the people within it? Often the best solutions aren’t simple outcomes; they’re winding, organic journeys where we listen, understand, create, collaborate and iterate together.

There’s no denying that AI is the opening theme tune to 2023. Debate is raging across the digital space, from design to content; from education to publishing. You can barely scroll for 30 seconds on LinkedIn without a post about AI or ChatGPT. It’s intoxicating. There are already countless examples where companies are saving time and money with AI: just look at this truly brilliant use of AI-generated conversation summaries from Intercom.

AI content and imagery: quickly becoming the norm

We’ve all seen how ChatGPT can perform all kinds of content-related tasks, from desk research and full first drafts of articles to rewriting SEO-friendly titles and suggesting alternative keywords to squeeze out that content juice.

Design’s new dawn is breaking in the same way. We can access conceptual imagery in seconds by entering prompts into a bot plugged into Discord. We can automate tasks as part of a design system, having trained a bot to understand how the components are structured and generate rapid UI mockups.

Despite triumphs like this, are more fundamental problems with Artificial Intelligence coming to light?

A quick solution isn’t always a good one

One consequence of these new sources of instant output is the democratisation of speedy problem solving. In this AI-enabled landscape, anyone and everyone can generate solutions that look like solutions and smell like solution but fail to address the problem at hand. In the words of anyone born before 1970, ‘buy cheap, buy twice’. You’ll often have to post-edit, or do the work again.

Could applying poor solutions to tricky business problems make matters worse? Would you trust an unqualified, anonymous, faceless person to make decisions about your personal life, or make healthcare recommendations for your children?

Whether AI-generated or not, we must always assess quality and trustworthiness against risk and cost to make the best recommendations for everyone involved.

AI doesn’t solve problems the way humans do

AI is only as good as the understanding of the problem at hand. A former colleague of ours once said that AI is better defined as Augmented Intelligence: an extension of our current intellectual abilities. It’s an interesting theory.

AI makes us think that output is what matters. And with it, the focus has shifted onto deliverables - or more accurately, the illusion of deliverables - with sometimes little thinking, meaning or value behind them.

AI offers up solutions to problems, but it can’t do this without being instructed by a human prompt. It can help us create rapid concepts, or multiple copy options, or present information in different layouts. But it can’t (yet) grasp the context in which these deliverables exist.

The AI debate shines a light on the need to understand problems better. There are things an AI can’t understand the nuances of, like:

The real value is in knowing how to identify, weigh and blend all these factors together to deliver quality results. If it sounds like hard work - that’s because it is. That’s the unique value that humans bring to the design process.

AI tools: bringing universal mediocrity to all disciplines

Another area AI struggles in: the intangible thing we like to call brand. We can of course ask it to create infinite things but it brings little to the table that hasn’t been seen a million times before. However, with the correct prompt and tighter request in can supercharge some elements of production; take these brand elements created using Dall-E.

However, brands need to pivot in new directions: towards the unexpected, inventive and unique, not the predictable absence of soul that’s become so common in recent years.

This is because of the way LLMs work: they’re only able to understand what has gone before, based on a limited dataset and the developers that defined it. It’s pretty unlikely your next cutting edge innovation or rebrand or repositioning idea is going to come from putting your hand in the lucky dip of the internet. Could an AI bot have made the jump between positioning Hewlett-Packard from a toner company to a printer manufacturer, or Netflix from video rental to streaming?

Where it does start to get exciting is at the point where future thinking, technical trends, and AI-based input combine with craft. This is where creativity is born: in a hot fusion of ideas, thoughts and learnings that bring something new and brilliant into existence.

Serendipity and lightbulb moments are the great leaps forward that keep brands alive - and humans are still firmly in the driving seat here.

Problems, problems, problems

Fighting the creep of banality, the temptation to oversimplify and the lure of easy fixes brings us back to the need for a deep, rigorous understanding of the problem space. This is counterbalanced by the very clear advantages of time saving it affords us giving us jumping off points or automating entire processes. The key is directing that time towards value creation. In other words, doing the hard mental work that make brands, well, brands - and not the keystrokes of prompts fired into a generation engine.

That’s what we love to do and it’s what keeps us coming to work every day - but that doesn’t stop us being a tiny bit jealous that Jasper doesn’t have to set his alarm for 7am on a Monday morning…

Related articles