It’s All About the Algorithms
We like to say our business is ‘all about the people’. But for how long?
I’d like to preface this by apologising if l come across as a Luddite. Protesting against the advancing tide of technology may seem a King Cnut-ish thing to do, but I (like most of us) come from a long line of Luddites.
In 2022, a friend asked me if I, as a copywriter, was worried by the prospect of my job being made redundant by AI. Then, my answer was ‘no’. Why? Well, a few weeks before, I’d been reading a piece of writing for a client and I, knowing the writer’s usual standards, became slightly concerned about their state of mind.
Image by Mohamed Nohassi on Unsplash
Happily, the fact that the writing was dull, poorly executed and lacking in insight or humour was not a cause for concern: quite the opposite, in fact. The writing, it transpired, was an experiment in AI. At the time, it was welcome evidence to me that AI was terrible, and my job was safe – for now.
Backend Open Hell
Fast-forward to 2024, when the sophistication and threat of AI has increased somewhat. I even use it myself occasionally, for transcribing interviews. It saves some time, but brings its own risks. For example: what does it mean if a live sound expert refers to ‘backend open hell’? It sounds like something from a monitor engineer’s fever dream. And why should we be careful of ‘a balloon sign’ in a theatre refurbishment? And why-oh-why was some unfortunate described as ‘poorly comfortable’?
Luckily, I know enough to decipher these AI gaffes: they are, in order – the ‘Wacken Open Air’ festival; ‘a building site’ (in Northern Ireland); and the Olivier Award-winning lighting designer ‘Paule Constable’. You have to keep your wits about you. And, quite frankly, that’s what AI doesn’t do – yet.
The mixing of metaphors is one thing that really grates on my goat, but another is so-called ‘customer service’. Automated call systems were just the start. Now, for many interactions, customer service has become a soul-withering waste of time. Falling into a loop with a chatbot when your query is not on its agenda causes a level of frustration that would make Kafka gulp. My last encounter with a chatbot ended with me asking it ‘Are you bloody joking?’. I knew it wasn’t.
Despite this being so damaging, countless companies see it as the way forward: because it saves money, it’s a winner. We hear a similar wholesale collapse of standards in voice-overs on social media videos. The stilted, inhuman noise that’s becoming increasingly common is no doubt cheap to produce, but it sounds like someone is mercilessly mocking the afflicted. And in a way, someone is: we’re afflicted by our acquiescence, and we’re being mocked.
Keeping The Bar Low
We’ve all heard the saying “no one ever lost money by underestimating the intelligence of the American public”. It’s based on a witty observation by the early 20th century journalist H.L. Mencken, and although he actually said “the taste” of the American people, his point still stands. Today’s equivalent might be that the entire developed world will put up with any old rubbish. If enough of us are happy to quietly accept appalling service and poor content, then that’s what we’ll be given.
In a way this is nothing new. With writing, as with any creative endeavour, doing it badly has always been an option. A company’s written content – its ‘tone of voice’ – is a key facet of its presentation to market. If you want your voice to sound like something produced by the Stepford Wives’ Copywriting Club, good luck to you.
“The greatest difficulty, it seems, is in training the essence of morality, the concept of right and wrong. It’s a bit like teaching a restless one-year-old, but with the added pressure of knowing that it will soon be running the world.”
I recently had a conversation with someone who is helping to ‘train’ AI – one of the army of people who are employed to advance and refine these tools. Working remotely, they give tasks to these machines, issuing feedback and corrections. They provide reams of references, conduct conversations and set exercises on every subject under the sun and beyond, teaching the finer points of good behaviour. These include how to be helpful, how to not try to ingratiate itself by pretending to experience human emotions, how to avoid ‘hallucinating’ (i.e. giving answers that it thinks are helpful, but are, well – made up). The greatest difficulty, it seems, is in training the essence of morality, the concept of right and wrong. It’s a bit like teaching a restless one-year-old, but with the added pressure of knowing that it will soon be running the world.
C-AI-veat Emptor
My hope is that people will recognise the risk of AI offering a Pyrrhic victory over the bottom line: you may save a few quid, but your customers and consumers might despise you. In our industry at least, it’s currently unthinkable. A technician who has to solve a problem 10 minutes from doors will not welcome an AI chatbot on the tech support line: they need a real person that they know and trust. And more generally, an industry that thrives on creativity and collaboration needs to retain that human element as long as it can. AI will play an important role, like the computing revolution has, but it has to support human creativity, not replace it.
When AI platform ChatGPT was used to write song lyrics in the ‘style’ of Nick Cave, the musician was not impressed. He said it was “fast-tracking the commodification of the human spirit by mechanizing the imagination”. He also said (less eloquently, but far more efficiently) that ChatGPT “sucks”. Would AI come up with that?
It’s all too possible that one day, no-one will be able to tell the difference between human and AI-generated output, or care very much about it one way or the other. Until then, let’s cling on to the unique, the unexpected, the human.
