As a copywriter by trade and a general advocate for human creativity, I’m probably in one of the main categories of people threatened by the stunning emergence of ChatGPT. I remember fielding questions from colleagues back in December, when the mainstream caught wind of OpenAI’s new chatbot. What did I think? Was I intimidated, curious, worried about my job? At first I indulged in a usual tendency to scoff. I am a humble writer with zero practical skills, but I refuse to be scared by the prospect of automated labour.
I’m not a tech junkie, and truth be told, I have never been that bowled over by artificial intelligence. I’d once seen a demonstration of a human conversing with an AI bot, and thought, “Wow, that’s impressive. How is that even possible?”. Then again, would I ever think the same thing when chatting to a human being who I happen to find quite boring?
Familiarising myself with ChatGPT, however, I have made discoveries that rocked my confidence a little. And yet, rather than shying away from the perceived problems it presents for writers, I still maintain that we needn’t be concerned. In fact, it can make our jobs a lot easier, which in turn only makes it easier to produce rich, engaging and original content.
ChatGPT is a powerful tool — but not for the reasons you might think
As writers, we shouldn’t be under any illusions about the capabilities of this new technology. ChatGPT is a marvel in that it can rapidly process the topics and details you ask it to and produce a wealth of substantial content. Some may find comfort in the fact that, at face value, what it generates is coherent, yet rather shallow. A robot may be able to assemble lots of information very quickly, but can it exercise judgement as a writer can?
Any copywriter will know that often the most stressful part of the job is familiarising themselves with areas they previously knew nothing about. Not only this, you have to do it quickly and somehow pull together various threads into a strong, rich and authoritative line of argument that competes with real experts.
A decent writer can fulfil the first part of the brief and produce something coherent and well-written. A competent one, on the other hand, adopts the perspective of the reader, consistently asking themselves the kind of questions they would expect to hear if they had to read it aloud to a real prospect. The difference between the two is what separates our job from the capabilities of a chatbot. Yet if a robot can do the first half of the job much quicker than we can, why not exploit it to the full?
The ideal research tool
One notable example of ChatGPT’s equally impressive and worrying capability was recently brought to light by a recruitment team, who unknowingly gave the AI platform a job interview after the bot completed a writing brief, ironically, on the secret of good writing. Its response was alarmingly good — clear, concise and convincing. Perhaps the most worrying part is that, at first glance, it contradicted my initial assumption that the bot cannot exercise judgement like a human can. The answer it gave was by and large a competent one that read well and felt true. That doesn’t mean it wasn’t incredibly dull.
While of course I have cause for bias, the purpose of writing to fulfil a brief is only one half of the job, and in fact, ChatGPT incentivises asking more precise questions — something which happens organically during the research process, even if organising your thoughts takes time. The beauty of the bot is that, as you are chatting to get a response, the more questions you ask prompts it to refine its answer. This way, you can get much more detail out of its search to generate swathes of information.
If we approach ChatGPT therefore as a starting tool for gathering necessary content to build a feature article, opinion piece or blog post, we can use it as a way to scaffold more detailed and precise hunt for knowledge that the writer can then synthesise. That way, we get the most well-informed background information for whatever we write, and it remains organic by following our line of questioning.
Turning briefly back to the problem of identifying a robot writing versus a human, although it may be a headache for teachers and recruiters, this is where another advantage of ChatGPT can be exploited. By giving the bot an instruction to take on a specific role, you can turn it into a combined researcher and interviewee when delving into a niche topic area by making it assume a clearly defined perspective.
This may take a few back-and-forth prompts, but as we’ve seen, its command of language gives it the potential to come up with some surprisingly humane responses that turn it into a kind of empathy-machine. A writer could then, theoretically, experiment with plugging in different ‘personalities’ as it were when exploring opposing points of view. This can be used to dig deeper into debate beyond the surface facts if you’re trying to find a fresh angle on a topic.
There’s a famous scene in I Robot that perhaps best encapsulates the dilemma of countering the idea of AI’s incredible potential with the suggestion that humans are inherently more creative. Will Smith’s Del Spooner, inherently distrusting of cold humanoid robots becoming an influence in the 2030s (not too far away folks), asks his machine counterpart, “can a robot write a symphony?”. The robot’s response, “can you?”, cuts straight to the point. Even if right now, AI can’t write an essay unprompted, this may not be the case forever.
Yet the rise of ChatGPT brings to mind the parallel question, “can a robot write a novel”. I haven’t abandoned my dreams to do so, but I remain convinced that the answer is a resounding ‘no’. Writing is not a logical process, and truth be told I don’t understand it half as much as I would like to. The purpose of writing is to provide clarity and judgement, cutting through a confusing array of information and helping the reader to understand something.
That judgement requires synthesising multiple points of view, and that process is not always logical but in fact, creative — and as The Atlantic suggested in a recent piece, for ChatGPT, creativity is all about the prompt. Harnessing artificial intelligence for creative writing for whichever audience relies on starting with a subjective angle which, for the time being at least, belongs solely to human beings. Until ChatGPT has been around the block a few times, I for one am not panicking.
However, neither should copywriters, customer support staff or examiners be complacent. Rather, I am eager to let it help save precious time — something a machine will never value quite as much as we do.