Why I don’t think AI is making writers redundant
Because a bot isn’t funny, original or thoughtful
Like most writers I occasionally worry that my livelihood may be under threat from AI. No day passes without it being in the news, most recently the launch this week of China’s DeepSeek.
Think of any straightforward assignment - what to visit in Barcelona, say - and AI can do the job perfectly competently. If it’s simply a list you’re after that may be all you’re looking for though even then you might want someone to put a personal spin on the information they’re giving you.
But judging by the success of Substack people are looking for connection with a real person. They - or at least I - don’t want what they read to be formulaic, culled from what’s already online. They/I want originality, attitude, humour, empathy - all qualities that the average bot lacks.
A trusted voice
The key to being followed as a food, wine or any other critic is becoming a trusted voice. One you welcome into your inbox because you know from experience they’ve proved reliable. If they rave about a wine, a restaurant or a place to stay you know it will be good or at least, if not to your taste, you’ll discover why.
A bot might well tell you what restaurants and wines other people rate - so could I albeit less quickly - but not give you a personal take on it.
It’s also about tone of voice. Good writing sounds like the author is talking to you. There’s a leaden clunkiness about most AI generated content that may provide answers but generates little joy. There’s pleasure in a well-turned sentence.
AI, so far as I am aware, doesn’t dig deep into a subject to find an original take on it or make random or quirky connections. It doesn’t reply to people’s feedback except in a polite formulaic way. You wouldn’t want to go out to lunch with it or look to it to assuage your anxiety. In fact it’s likely to make you more anxious.
I also wonder how reliable it is.
I set ChatGPT a couple of challenges.
First, ‘how do I make the Wagamama chicken katsu curry?’ (being a bit obsessed with katsu at the moment). It obligingly came up with a recipe for a ‘homemade Wagamama-style chicken katsu curry’. But where did the recipe come from? There are obviously plenty of katsu recipes out there but how did it know if it was similar to Wagamama’s which so far as I can see is not online elsewhere? AI can’t taste.
Then I asked it to write a wine review in my, Jane McQuitty’s and Jancis Robinson’s style all of which were pretty similar. ChatGPT has obviously been instructed that wine reviews need to be long and flowery. Mine read like this.