AI: clever but not clever enough
Ashgrove Marketing’s Terry van Rhyn puts his mind to the topic of artificial intelligence and whether machines can ever be truly creative
I have been bombarded recently by articles, talks, programmes and many enthusiastic personal recommendations about the wonders of artificial intelligence and how it can improve productivity at my agency.
Many times I’ve been told to simply type in any topic you like and within seconds this AI bot will write you a report, opinion piece, content or post. So, are the days of sitting and staring at a blank Word doc screen trying to find the words to get a story across gone forever? Based on my current experience, I’d have to say absolutely not!
You may call me old school, not progressive, resistant to adapt to new ideas or perhaps simply old. All possibly true but hear me out.
As a test, I entered a question about why clients should use Ashgrove Marketing for branding purposes. I was in awe at the eyewatering speed at which this AI bot was able to spit back an 800-word article. But I wasn’t impressed for long as I read through the piece and realised how superficial and repetitive it was, and void of personality or a consistent tone of voice.
It used words and phrases that we have been known to say but not in any meaningful way. There was no passion, no quirky ideas, nothing that would make us stand out against any number of other agencies.
Initially, I was ready to embrace the idea of AI being a great research tool to help speed up the intelligence gathering we often require as background to developing our brand strategies. And it may still be useful for this purpose.
But as for actually developing new content, there is a long way to go. And there is another, potentially more tricky and expensive issue with AI-generated content.
As I dug deeper, I began to question how it was even legal. If the AI bot is crawling around the web searching and pulling data from other relevant articles or from content on websites – are we now crossing into plagiarism or copyright territory?
It’s mooted that as the content is generated by a machine, it’s immune to copyright infringement or fines for failure to give proper attribution. Yet the human-generated content it is working from may very well have legal protection.
Technology site The Verge recently featured the case of Hollie Mengert, a Disney illustrator who found that her art style had been used in an AI experiment by an engineering student in Canada. The student trained a machine learning model to reproduce her style and then released a version of it for anyone to use. He even attached Mengert’s name to the description of it – all without her permission or even her knowledge. Understandably she was not impressed.
As with any disruptive technology, we will have to go along on this journey to discover how AI actually fits together with current regulations and laws. Like most new things, it will have to go through a natural cycle of scrutiny before it can be accepted as the norm.
I remember well a similar disruption to the design and advertising world in the 1980s. In those days creative studios employed artists, most of whom studied art and design or commercial art as it was known back then. Creative concepts were presented to clients in the form of elaborate “scamps” or sketches to convey and visualise our ideas.
These could take days and weeks to create, and were designed to sell a concept through exceptional visual representations of the idea. We used Letraset, repro cameras, Pantone marker felt pens, Gouache paints and specialist typesetters to refine these pieces of art. The client would then approve ad campaigns worth hundreds, if not millions, based on these concept sketches!
Enter Steve Jobs and the Apple Mac which disrupted an entire industry. Suddenly, we were able to present concept ideas in a slick format with actual images and a variety of copy fonts with different size options which often looked like the final product.
The clients were in heaven as they could easily visualise the idea and also ask to increase the size of the logo without it taking days for a frustrated artist to redo the scamp! The result was that we no longer were employing artists who understood colour, layout and composition, but rather technicians who had the technical skills to operate these Macs.
Those artists who could not adapt had a hard time, but it opened a door to a new breed of artist who did not have to go to college or university to get paint and charcoal on their fingers (and clothes), or walk around with an art portfolio the size of a small car.
Now traditional repro houses no longer exist, and you can buy sheets of Letraset on eBay for nostalgia. Most artists came around to operating Macs eventually, while others remained relevant by becoming specialist freelance suppliers of original art to the ad agencies. Even today, with the many photography and art libraries around we still require at times something original that is completely unique.
And that might be the Achilles heel of AI – finding something magical and unique for a brand message or a sparkling page of copy that sets it apart from the competition. Not just a regurgitation of what’s already there.
To my mind, the jury is still very much out on AI. Can a machine programmed to learn ever be really creative or “think outside the box”? It may have its uses – and some of these in turn may prompt creativity in its users – but often you just need the crazy human brain to make some random and unexpected connections to set your brand or campaign alight.