I love seeing typos.
Much as I adore ChatGPT (OpenAI) for its ideas, refining my clumsier thoughts, proofreading short texts, finding emojis and providing references, I’m also noticing its limitations.
At times, it can be misleading, and it lies like a cheap watch. Recently, it has offered me some dubious responses, reminiscent of my student days in the early 90s, when I used to invent book titles and authors for my essay references (there was no internet then – how could anyone possibly know?!).
ChatGPT is repetitive
I’ve started to notice ChatGPT’s highly distinctive words and writing structures.
Words like ‘foster’, ‘additionally’, ‘furthermore’, ‘sumptuous’, ‘moreover’, and others.
The longer I use the platform, the more I notice these linguistic patterns making their way into people’s social media posts, articles, and even job ads. There are also conspicuous tell-tale signs, such as the Americanization (deliberate use of the letter ‘z’ here!) of words, excessive use of colons in bullet points, and paragraphs that seem to lack soul.
Perhaps the gravest offense of all is accidentally incorporating ‘regenerate response’ at the end of a post (yep – it happened!).
Embrace the mistakes!
In my junior marketing career, typos and mistakes were a constant fear, as they gave off an aura of amateurism and incompetence.
During my time in advertising, I certainly had my fair share of disasters. I think that enough time has elapsed for me to finally forgive myself and hope that others have forgiven me.
One memorable calamity was a missed digit in a telephone number for a direct response ad in News of the World – I swear I proofread the ad at least ten times. Thank you to my then Account Director Justin for handling the situation with the client ‘discreetly’.
Other incidents involved signing off a huge poster in a busy tube station with a typo in the first sentence, and the unfortunate sending of the wrong version of an ad to the Daily Telegraph, leading to a fiery telling off from a senior marketing manager at an investment bank in the City.
Fortunately, such errors were infrequent, but they still haunt my nights, causing me to wake up in cold sweats.
Keep your accent
ChatGPT has inadvertently stripped away our authenticity, leaving us at risk of losing our unique voices, genuine mistakes, and individual quirks. There’s a danger of everyone gravitating towards a standardised way of writing, swapping their regional accents in favour of Received Pronunciation (RP) or the Queen’s English – should that be the King’s English now?
So now, in 2023, when I come across a typo in someone’s post, I love it. It serves as a testament to their authenticity. Just as Canva may have diluted the creativity of visuals with repetitive carousels, ChatGPT seems to be homogenising our language, washing away the colourful richness of our clumsy human words.
I have a vivid memory of a chance encounter at a rave in 90s London. Amidst the pulsating music and dim lights, I bumped into a loved up sweaty stranger wearing an oxygen mask who unexpectedly embraced me and uttered the words, ‘Use, don’t abuse.’ I couldn’t help but assume he was referring to illegal substances, perhaps drawing parallels to the infamous ‘Grange Hill’ episode featuring Zammo and the powerful message of ‘Just Say No.’
This is the way we should embrace ChatGPT – provide it with our outlines, draw inspiration from its ideas, and then move on – hence the ‘use, don’t abuse.’
Embrace your clumsiness, for there’s a charm in it that I adore. Let’s avoid monotony and plainness – give me something vibrant, messy and distinct any day.
Let’s make mistakes
As we venture into this new era of accessible AI, we must recognise that it is a learning process on both sides. ChatGPT is still unfamiliar with our wonderfully messy and original human ways.
Let’s never lose that – it’s precious.
As Coco Chanel said, ‘The most courageous act is still to think for yourself. Aloud.’