ChatGPT – the perfect apprentice for the data storyteller

There’s a whole lot of panic going on about ChatGPT, particularly among those who write for a living. Those who write copy, those who write strategies, and those who write training courses. There’s a sudden, ice-in-the-bowels terror among purveyors of HI or Human Intelligence that OpenAI’s Generative Pre-training Transformer will soon put them out of business. For generations, the threat of technology to knowledge economy workers seemed remote, but in an apparent blink of an eye it appears as though computers are now able to understand and respond to questions we ask with structure, authority, and aplomb.

In the spirit of Mark Twain’s widely misattributed aphorism, I sense that the reports of the death of the careers of ‘them what write’ has been greatly exaggerated. Successive generations of innovators always stand on the shoulders of giants from previous generations and the history of human advancement shows that technological development doesn’t put us out of work. Indeed, populations have only been able to grow exponentially because of the invention and channelling of tools and techniques that remove drudgery and make old ways of working obsolete.

Standing on the shoulders of giants

From fire to electricity, from mining and refining of everything from iron ore to rare earth metals, from the abacus to the supercomputer, progress is predicated on the next generation not having to do or discover what their predecessors did. Without microscopy, Pasteur, and Crick and Watson, Sarah Gilbert and her team wouldn’t have been able to develop an effective COVID vaccine and get it ready for clinical trials in just 65 days. Humans don’t just thrive and advance because of genetic success; it is memetic prowess – the ability to pass intellectual discovery from anyone to anyone else – that is our revolutionary superpower.

“Fair enough,” say the worried writers, “but ChatGPT is different. If you can simply type ‘Write me a 600-word article for LinkedIn on challenges facing property developers – please’ and refine it with ‘Make it specific for the continental European market’, ‘Add in some humour’, and ‘Create ten tweets with SEO-optimised hashtags’, what role is there for people like us?”

The sorcerer’s apprentice

For me, ChatGPT is – even in its current, early, nascent form – a brilliant sorcerer’s apprentice. It’s got knowledge – data – at least up to the end of 2021. By accessing that data, it can readily build structure into its responses, because previous articles for LinkedIn, strategy papers, or strings of tweets have a format, and it can quickly access what that format should be. And not only is it data rich, but it also has genuine storytelling smarts. It’s not only a sorcerer’s apprentice; it’s the ideal data storyteller’s apprentice, combining the often fire-and-ice worlds of narrative and numbers, stories and statistics. And as that’s what I and my team do all day, you might think I ought to be worried.

In fact, we couldn’t be happier with this apparently sudden breakthrough. As a means of generating ideas, formats, and usually well-expressed thought starters, ChatGPT is just brilliant. And that’s from the current version, likely to be superseded by GPT-4 sometime this year – and even this quarter, if Microsoft’s $10bn-plus investment in the technology and platform has anything to do with it. But it doesn’t remove the need for the application of HI to AI. It makes it ever-more vital.

While communications agencies might subcontract the creation of first drafts of news releases, blogs, or LinkedIn posts to ChatGPT and other, soon-to-launch rivals, the need for tailoring to specific client circumstances, fact-checking, introducing topical references, sprinkling appropriate humour or seriousness – all of these skills and more require levels of nuance, subtlety, and wit that I and many other non-doom-mongers believe will remain distinctly human for some time yet.

Data + story

Let me give you an example. I was talking to a friend who works in market research recently. He told me that – as an experiment – he recently asked a ten-years-qualified team member to write a questionnaire and discussion guide for a new booze brand targeting 20-34 year-old urbanites in Latin America. His colleague – one of the best at this task in the agency – took a day-and-a-half to come up with 30 questions and a logic flow that passed muster. Lots of things, from client emergencies to team meetings, got in the way. As a comparator, he asked ChatGPT to do the same. With half-a-dozen refining requests submitted to platform, he then passed his five minutes’ work to a graduate trainee, gave her a couple of previous questionnaires for the same client as inspiration, and asked her to refine the AI-generated output. In 90 minutes, the AI x graduate output was at least as good as the ten-year veteran’s. It was also rather better structured.

Now this doesn’t mean that ChatGPT has put the veteran out of a job. Far from it. But, with intelligent human curation, it can help to take the drudgery out of a routine part of the job and free-up time for subsequent analysis and surfacing and articulating genuine insights into the data generated by the research. The kind of thing the veteran should be doing to add real value to the research output, rather than the more mundane task of questionnaire writing.

The spirit of the age

The defining equation for success in the modern knowledge economy is “Analytics + Storytelling = Influence”, and ChatGPT is that equation writ large. We shouldn’t worry or cede total control of what we write – whatever we write – to this precious talent. We should stand on its shoulders – and the shoulders of the giants whose intelligence it can mine in seconds – and then review it and incorporate it in ways that allow us to be smarter yet. By combining data and narrative skills, ChatGPT is a tool for the age, the very essence of the Zeitgeist.

Note: none of this blog was written using ChatGPT.