Amid growing conversations around AI-generated content, FashioNZ stares automation in the face and asks what we are willing to accept from our fashion media
Last week, a chill went down the collective spine of Aotearoa’s fashion and media industry as it was announced that one of the most experienced and highly-regarded editors in our orbit would be stepping down from her role, as part of our national news publisher’s latest streamlining efforts.
For 15 years, Amanda Linnell has steered NZ Herald’s Viva through recession, redesign, the launch of its phenomenally successful website, social media domination, pandemic pivots, shifting editorial priorities (hello, commercial publishing), and multiple rounds of downsizing. All while pumping out innovative, intelligent, inclusive, industry-leading content for daily digital, weekly newspaper and quarterly magazine deadlines.
As this news broke, I was miserably plodding through week two of a cold that wasn’t getting any better, watching my own daily digital deadlines slip through the net with an alarming level of disaffectedness for someone whose own job has always been (and still is) about as stable as a stiletto on wet cobblestones. My therapist would call this growth; I’d probably just call it cough medicine.
Maybe my publisher was on the extra-strength Lemsip, too, because by Friday, when I hadn’t been able to deliver on a week’s worth of content, much less get a handle on the next week’s, he said, ‘Why don’t we get ChatGPT to write it?’
And so, being the charlatans post-modernists that we are, we began inputting prompts into the free software – ‘Kate Sylvester female muses art and literature’; ‘Ruby community-building’; ‘Trelise Cooper top five scandals’ – and seeing what it spat out.
It wasn’t about cost cutting. It wasn’t even about outrage bait. At the time, it was simply about exploring the available options for saving staff health and sanity. In my opinion, this came at the price of quality and nuance but (also in my opinion) this nuance will be lost on many casual readers. By which I mean, it’s the strength or weakness of our own literacy and reading comprehension that determines the usefulness of AI-generated content.
Becoming more invested in the experiment, we began to test AI’s capacity to create editorial imagery – by far the most expensive item on the editor’s line sheet. In the leanest of times with the meanest of budgets, a cover shoot for a quarterly glossy magazine in New Zealand in 2023 might cost between $5,000 and $10,000. That money is quickly gobbled up by photographers and stylists and their assistants’ fees, hair and makeup, location or studio costs, extra lighting and tech, secondary photography (i.e. videography, film or behind-the-scenes content), model and agency fees, couriering clothing samples across the city, insurance, catering, parking and retouching.
Not to mention the days, if not weeks of ideation and coordination that go into locking down a concept and a production team, as well as the invisible cost of having an editor and/or creative director, fashion assistant and possibly a journalist on-site throughout the day, and the several more hours of post-production involved in image selects, layout design and writing cover lines. Far from something that is just slapped together in seconds, every frame, every millimetre, every Pantone is considered and debated and decided on for a reason. That reason being to sell magazines.
It is not self-aggrandising to say that what the consumer ends up paying for is an artwork. A fashion magazine is a piece of decor that you can display in your home as a signifier of taste and status. Just look at your own Instagram feed for proof.
It’s become popular to sneer at those who appear to buy books and magazines not to read, but to use as props for social media. But this attitude overlooks the historic function of books as status signifiers. Before the printing press, books were such valuable objects that they were the exclusive domain of monarchs and religious leaders.
Post-printing press, when the cost of materials was high and literacy was low, they continued to be elite possessions passed down as heirlooms along with jewellery and textiles and other decorative arts. Mass publishing and widespread literacy changed this, but as inflation causes the price of printing materials to skyrocket again, and as our modern definition of literacy shifts, we have come full circle.
Except now, for zero cost and a time investment of about 1.9 seconds, AI can generate an image of an ‘attractive, Māori woman wearing a Karen Walker dress, standing against a backdrop of New Zealand natural scenery’ that would get a pass in a high school graphic design class. Spend a little more time and money and you will wind up with something much more believable. Later this week, we speak to an Auckland-based AI expert about the possibilities and limitations of computer-generated cover art, and how it feels as a highly-specialised creative to have the tools to cannibalise the jobs of other highly-specialised creatives.
Because these roles are highly specialised, which is why the hubbing of editorial teams by large media corporations is so disturbing. You would not expect your neurologist or your plastic surgeon to perform the role of your GP except in an apocalypse. They might have started in the same place, and they both play equally important roles in the medical ecosystem, but across decades of intensive training they have established wildly different skill sets. One knows a lot about a little, the other knows a little about a lot. It’s understandable why penny-pinching publishers want us all to be the latter, but you need depth, not breadth to cure the brain rot of a society desensitised by a deluge of unfiltered information.
Incidentally, I am writing all of this in a Google Doc that is persistently trying to autocomplete my sentences. It has been wrong every time.
As well as the obvious limits of artificial ‘intelligence’, what occurs to me as I reject each new suggestion is that completing someone else’s sentences is a symptom of ADHD – something that many creative people reading this will know. A common fear of people on the ADHD diagnosis pathway (or indeed, considering any psychiatric medication including those used to treat the depression and anxiety that is also prevalent in creative people) is whether their creativity will be blunted.
Like a digital amphetamine, is AI going to be responsible for a mass blunting of creativity, assisting in the rapid churning out of generic and bloodless automated content? Or could it actually take the pressure off hyperactive, attention-deficit, overworked and underpaid journalists and creatives, enabling them to focus their energies on executing high-value work, with more artistic integrity than what the current situation allows them the mental space to produce?
What safeguards could be in place for employees so that they could be the first to benefit from the potential of AI, and not live in fear of its making their roles redundant? What if AI itself was the safeguard; the solution to providing sick cover that doesn’t create extra stress for other team members?
It’s a utopian answer to a dystopian situation that creeps ever-closer, as each new round of lay-offs stretches the creative and physical capacities of the last of our highly-skilled editorial workforce.
And it’s not a definitive answer. As an editor reviewing what my ‘team’ of robots produced for me this week, I would send it all back. Not for the factual inaccuracies (it pains me to say that AI’s grasp of grammar and narrative structure is better than many professional writers), but for the lack of heart and humanity.
If I wanted to know the basic plot points of a designer’s career give-or-take a few un-factchecked claims, I would consult their Wikipedia, but I don’t. I want the insights and vulnerabilities that only a seasoned interviewer with empathy and excellent ego-management can draw out of a subject with 45 minutes of genuine conversation, a measured amount of flattery and, occasionally, a little bit of wine.
For now, the result is going to be more engaging than what a highly-trained computer can achieve. But you have the articles in front of you. Have a read and then ask yourself, could we live with this?