Humanity can save your job from ChatGPT
Panic is spreading in what many of us regard as the world’s most important industry. For decades, editors have been adding columns in a bid to explain the hidden meaning of the whirligig of daily events. Now we columnists must confront the possibility that ChatGPT will be able to do our job in a fraction of time and without all the fuss and expense.
Matthew Parris, one of Britain’s finest columnists, recently admitted that his assistant had asked ChatGPT to produce a Parris column on Sir Keir Starmer. “The result is scary,” he admits. “The column goes through the options in a perfectly coherent way, and a reader could easily conclude the work was mine, but submitted on a dull day...” Since that dull day, the people behind Chat GPT, Open AI, have released a major upgrade in the form of GPT-4. The original ChatGPT only scored in the 10th percentile on the bar exam while GPT-4 passed in the 90th percentile. Has Parris improved as much as his imitator?
It’s easy to make fun of us nervous hacks. Will neoliberal columnists who have spent decades advising coalminers to retrain as computer programmers remain as enthusiastic about creative destruction now that they are the target? Will Left liberals who celebrate the joys of open borders remain so generous when they face a future as cooks and bottle-washers? Perhaps it’s time for the opinion classes to issue a collective apology to the Luddites — the obsolete handloom weavers and stocking-makers whom they have routinely and lazily mocked for smashing machinery rather than leaping on the locomotive of history.
But remember that columnists are human, too. Sack them and they weep real tears. And remember, too, that if ChatGPT can replace columnists, it can replace all sorts of people who try to spin words and numbers into argument and analysis. You will be next.
Can anything stop the digital destruction? I think so. At bottom, ChatGPT is no more than a vast recycling machine. It can search our collective digital brain for precooked ideas and preassembled facts, and then churn them out as columns. It can imitate whatever style you tell it to imitate. But it cannot provide the human element — vivid observations or fresh ideas or leaps of imagination. The best way for columnists to survive in the world of artificial intelligence is to write more human columns. This general rule — avoid destruction by upping the human element in what you do — applies to most other knowledge-intensive jobs.
Here are some ideas for avoiding the coming “columnageddon”:
• Live an interesting life
Most columnists these days proceed from school to university to the newsroom without pausing to get their boots dirty. This immerses them in a world of experiences and opinions that ChatGPT can easily recycle. Great columnists have rich personalities formed by a cacophony of experiences. Winston Churchill learnt how to write journalism not by going to journalism school — or indeed to university — but by getting himself sent to war zones and sending dispatches to whatever newspapers would pay the most. Frank Johnson left school at 16 and got his first reporting job in journalism on the North-West Evening Mail, where he had to make the minutiae of local life interesting. This gave him an eye for detail and a sense of absurdity that produced some of the best political columns of his time.
A subdivision of the interesting life is the tragic life: artificial intelligence, being rational, cannot reproduce the effects of slowly destroying your life by the persistent application of obscene quantities of alcohol, tobacco or other noxious toxins. Jeffrey Bernard wrote his “Low Life” column in The Spectator for 20 years celebrating his love affair with the bottle and the geniuses and bores who inhabited the Soho of his day: “a suicide note in weekly instalments”, as it was once dubbed.
Another legendary drinker, Henry Fairlie, made an immortal contribution to British journalism by inventing the idea of “the establishment” — the interconnected group of families and institutions that run the country regardless of changing governments. But he was obliged to flee the country because of his habit of borrowing money without repaying it and offending everybody in power, particularly his employers. He ended his life living in a small room in the Washington office of The New Republic, where his few possessions included the complete works of Dickens.
• Cultivate ever-changing hinterlands
It is possible to imagine ChatGPT producing a mélange of Roger Scruton’s High Tory beliefs — tradition good/modernism disgusting. But Scruton brilliantly illustrated his beliefs by encomiums to fox hunting, fine wine and Wagner. It is hard to imagine even the cleverest algorithm conveying a sense of what it’s like to fall off a horse in pursuit of a terrified fox or sitting in Bayreuth’s deliberately uncomfortable seats listening to Norse gods and heroes singing to each other.
But having a few hinterlands is not enough: you need to keep acquiring new ones, lest AI learns to imitate them. The great management guru Peter Drucker recommended that people should devote time every day to studying new subjects that were completely unrelated to their daily lives (when I visited him in his ranch house in Claremont, Southern California, in the late 1990s, he was deeply immersed in tenth-century European history).
• Generate new knowledge
A worrying number of columnists spend their days doing what AI can do better and faster: searching the internet. The only salvation in the long term is to put new stuff into the machine rather than extracting old stuff. There are two ways of doing this.
One way is to immerse yourself in the real world: interview new people (and observe their peculiarities) or visit new places (and smell the air) or stroll the factory floor. The best political columnists spend their lives mixing with politicians — and some like it so much that they cross over, as Alastair Campbell did with Tony Blair and James Forsyth has done with Rishi Sunak. Rowland Evans and Robert Novak pledged that every column would contain at least one political revelation, an achievement that was made possible by their habit of palling around with the powerful. John F. Kennedy had his first dinner as president-elect with Evans.
The second way is to think fresh thoughts. This involves not just immersing yourself in the world of thought but also mixing and matching ideas from different intellectual traditions. Rupert Pennant-Rea, the editor-in-chief of The Economist from 1986 to 1993, said that what really excited him were “scoops of the mind”. George Will avers that the columnists who most impress him are those who write about the “inside of public matters: not what is secret, but what is latent, the kernel of principle and other significance that exists, recognised or not, ‘inside’ events, policies and manners”.
These two forms of originality are not mutually exclusive.
• Avoid predictability
AI thrives on predictable patterns in opinion or rhetoric: it is easy to imagine ChatGPT producing the “inverted pyramid of piffle” that is a Boris Johnson column. The only way to survive is to keep surprising us. If you think Trump is the Devil, try getting inside the skin of a Trump supporter; if you’re an anti-wokist, try seeing what the world looks like to a struggling trans teenager.
Some of the best columnists embrace a host of mutually incompatible beliefs and struggle to reconcile them: Andrew Sullivan is a conservative, gay Catholic who has variously idealised George Bush and Barack Obama, and is now as fierce in his condemnation of wokery as he was once in his support of gay marriage. Try that, ChatGPT!
Others have relished breaking with their tribal groups, a quality that only grows in value as opinion, like everything else, becomes more tribalised. George Orwell was a lifelong socialist who relished offending his congregation. “It is a strange fact, but it is unquestionably true, that almost any English intellectual would feel more ashamed of standing to attention during God Save the King than of stealing from a poor box,” was typical of his intellectual-baiting. Christopher Hitchens was a leftist who ended up supporting the Iraq war and raging against “Islamofascism”.
• Most important of all: be funny
Humour seems to be the most inimitably human of all attributes. Certainly, AI’s attempts to be funny have thus far been dismal. It is also the acme of the columnist’s art. One historian of journalism suggests that the “pasquinade seems the forerunner of the column” — a pasquinade being a publicly posted piece of satirical writing named after Pasquino, a Roman tailor, who wrote lampoons of his city’s wealthy and powerful citizens. This thumbing of noses is essential to the columnist’s craft. Some of the earliest columnists such as Ambrose Bierce and Will Rogers were primarily humorists — although the humorous column seems to have died in the land of its birth. Humorous columns also last longer than most serious ones.
No one these days reads Bernard Levin, who once bestrode the columnular world like a colossus. But Auberon Waugh, Evelyn’s son and a brilliant satirist in his own right, has his own Twitter feed that pumps out bits of columns from decades ago. Here is Waugh on the mass murderer Dennis Nilsen:
“Like so many social and emotional cripples he drifted into left-wing politics and became branch organiser of the Civil Service Union, but even there his boring self-righteousness failed to secure the warmth and comradeship which he sought. So ... he was reduced to seeking the company of corpses as being the only people who would not walk out on him.”
It is hard to imagine ChatGPT generating any of this, let alone allowing it to bypass the “woke check” that is being installed in computers alongside the spell check and the grammar check.
One of the most powerful arguments in favour of automation is that it forces producers to be more productive by focusing on their comparative advantage. This will be surely the case with columnists. ChatGPT can do all the predictable things — producing anti or pro-Trump rants or summarising the causes of the war in Ukraine. But can it change your mind? Can it produce “colour” that encapsulates an historical moment? Can it persuade a politician to spill the beans? Can it generate a new way of looking at the world? And can it make you laugh? Not yet — and, in some cases, not ever.
What applies to columnists applies to all knowledge workers. Some companies are demanding that workers produce more content more quickly to keep up with AI: translators who were once required to produce 1,000 to 2,000 words a day are now, in the world of Google Translate and other aids, expected to produce 7,000 words. This modern times-style speeding-up is a road to nowhere. Humans cannot compete with AI when it comes to speed in recognising patterns in vast bodies of data or processing mere words. Computers are taking over all sorts of jobs from preparing legal briefs to interpreting X-rays to analysing the relative performance of companies. The answer to this challenge is not to destroy the chatting machines, tempting though that is, but to rediscover what makes us human.
We need to immerse ourselves in the world rather than communing with computers all day: AI cannot feel the direction of the wind changing or sense the atmosphere of a shareholder meeting. We need to cultivate sources in the human world over dinner and a drink: chatbots cannot get people drunk and wheedle out indiscretions. We need to relish the imperfections of the human species such as its talent for humour, which is surely tinged with sadism, and its powers of creativity, which are surely connected with our ability to lie.
The great source of solace in a world in which computers are getting cleverer all the time is that the only thing that can save us from redundancy — our imperfection — is here to stay.
• Adrian Wooldridge is the global business columnist for Bloomberg Opinion. A former writer at The Economist, he is author, most recently, of The Aristocracy of Talent: How Meritocracy Made the Modern World
2. Please respect the use of this community forum and its users.
3. Any poster that insults, threatens or verbally abuses another member, uses defamatory language, or deliberately disrupts discussions will be banned.
4. Users who violate the Terms of Service or any commenting rules will be banned.
5. Please stay on topic. "Trolling" to incite emotional responses and disrupt conversations will be deleted.
6. To understand further what is and isn't allowed and the actions we may take, please read our Terms of Service