Post #57
June 1, 2022
Tamara O’Brien
FW copywriter Tamara O’Brien considers the challenges and opportunities of writing and design by machine.
Oh, it’s grand being a UK creative. Not only are you licensed to use your imagination in the commercial world – thus achieving the holy grail of art with remuneration – your trade is pretty much immune from the effects of offshoring, globalisation and technological advances.
Because nothing and nobody can match Team GB’s design, media and music industries for flair and originality, right? ‘You can’t fake creativity’ has been the mantra for most of my copywriting career. It’s the USP of an industry worth billions to the UK economy[1], and a good living to the individuals working in it.
But wait, what’s that coming over the hill… artificial intelligence (AI) that can generate art, design, copy, poetry, editorial and the like, that’s indistinguishable from the real thing? British advertising’s stock-in-trade of wry humour and wide scope of reference, wrapped up in playful language – a machine can do that now?
Excuse me a minute while I sink to my knees Charles-Heston style. “You finally did it—you maniacs! You blew it up! Damn you all to hell.”
There, that’s better. A bit.
Deepfake it til you make it
What triggered this realisation was a fascinating article by Sean Thomas in The Spectator (23 April 2022) about the latest generation of AI.
Originally backed by Elon Musk and Microsoft, OpenAI is a company that aims to bring about a human-friendly, safe path to AI. In 2018 it launched its Generative Pre-trained Transformer, or GPT. The methodology was simple: cram a fancy computer with immense amounts of random information, in the hope it will eventually generate artificial general intelligence (AGI).
And that would change everything, because AGI is the closest a machine can get to human consciousness. Beyond mere AI, AGI is a machine independently deploying intelligence to plan, learn, solve problems, use what’s known as ‘commonsense knowledge and reasoning’, communicate in natural language – and integrate all these skills in a human-like manner.
The report that writes itself
We thought we wouldn’t get to this point for decades. But it turns out that the 2020 version of the model, GPT-3, the latest at the time of writing, is pretty much there. Like some monstrous copywriter, it’s been stuffed with everything on the internet, and its increased processing power – ‘the biggest neural network on the planet’ – has made it capable of producing anything from blog posts to music to short stories in the style of Jerome K. Jerome.
It works on the principle of autocomplete, riffing on word prompts to produce content that’s logical and sounds human. At the moment, it can’t go much beyond a 300-500-word article without descending into gibberish. But the time is coming when a journalist or copywriter just needs to provide a heading and subheading, and GPT-3 will do the rest. Imagine that when you’re struggling with your governance review.
Picture this
Designers needn’t think they’ve swerved the AGI juggernaut either. Thoughtfully, OpenAI has developed a graphics companion to GPT-3, called DALL-E. It all started when they asked GPT-3 to create images from language prompts (the brief, as we humans say). These include such PowerPoint favourites as ‘an isometric view of a sad capybara’ and ‘a baby daikon radish in a tutu walking a dog’.
And it did so with aplomb. DALL-E can create images of realistic objects (‘a stained glass window with an image of a blue strawberry’), or objects that do not exist in reality (‘a cube with the texture of a porcupine’). Just feed in the parameters and out pops something cute or believable. Designers wishing to give themselves a sleepless night can view the results on the DALL-E website.
Where do we go from here?
In his Spectator piece, Sean Thomas does not see a bright future for the graphics industry. He concludes ‘Basically, don’t send your kids to art college. Because DALL-E is possibly going to destroy many of the commercial art jobs in the world.’
He’s speaking from a journalistic perspective, of course. Actually, I’m more optimistic about creativity, certainly in the world of corporate reporting. Because here, writers and designers create their own briefs (signed off by our clients, of course!). Our role isn’t about responding to a clearly defined direction. We work together to help companies judge what they do and don’t need to communicate, and the best way to express it – changing and refining these judgements as the company story unfolds across past, present and future. In my 30 years in the business, no two client briefs have ever been the same – even for the same client.
Even beyond reporting, there’s more to creativity than coming up with new combinations of the same source material of words and pictures – even if the source is everything on the internet. Because humans do what AGI can’t: and that is to create our own prompts, from the haphazard consequences of living, feeling and perceiving in the real world. Chance encounters, inefficient but intellectually stimulating tangents, accidents of language and science, the wayward aspects of cognition – human beings delight in such phenomena, and as creatives, we infuse our work with them.
AGI could help with the heavy lifting of corporate storytelling, for example in assembling a mass of disparate information from multiple sources, and presenting it clearly and logically. That would be nice. But the subtleties of meaning, texture and interpretation? These will always, I hope, come from ‘wetware’ like you and me.
[1] In 2019, the Department for Culture, Media and Sport estimated that the creative industries contributed £115.9 billion to the UK, accounting for 5.9% of the UK economy.