Here’s a top tip if you want to woo the love of your life this Valentine’s: don’t use AI to write your card. Large language models might have mastered the art of summarising a report, creating consistent customer comms or crafting inoffensive LinkedIn posts, but matters of the heart? Not so much.
This week, I asked CoPilot to write a Valentine’s message. Here’s what it gave me: “Thank you for being my partner, my calm, and my joy. I love you more every day.” If I wrote that in a card to my husband, I genuinely think he’d ask if I’d had a knock to the head. But I can’t blame the LLM entirely…
Shall I compare thee to my training data?
Sure, AI can draw on every love poem, sonnet and letter it’s ever trained on, but it doesn’t know if I call my husband Snuggle Bunny (I don’t) or if I’m more likely to praise his roast potatoes (not a euphemism) than say something soppy. Without a detailed prompt, including pet names, in-jokes and examples of how we normally write to each other, all AI can do is fall back on romantic clichés. I didn’t fancy sharing all that info with our robot overlords, so Copilot didn’t have the context it needed to do the job. That’s not the only problem…
Roses are red, violets are blue, I know this rhyme wasn’t from you
There are loads of times when people really don’t mind reading AI content. Customer service chatbots and agents are a great example. They often save us time, and we don’t think the companies using them are trying to catch us out or pretend their bot’s a person. But boy is the opposite true if you’re using AI for something more personal.
Readers of Reddit AITA threads have probably seen countless posts from heartbroken souls, furious that the love of their life has written cards, letters and even wedding vows with AI. The people caught using AI argue that they’re not good with words and needed the help. The ones on the receiving end call out their (usually now ex) lovers as lazy and inauthentic.
And that’s the crux of most conversations we have in our Intentional AI training sessions: when does using AI make sense and when should humans keep hold of the pen? Sometimes the answer feels obvious. When it doesn’t, ask:
- Why am I using AI for this?
- How would the person reading the message feel if they knew AI wrote it?
- How would I feel if someone called me out for using AI here?
If your message is functional, with content you can easily put in a prompt, and your goals include saving time or keeping a consistent tone across lots of messages, then using AI makes sense.
If you’re writing something you think only you can say – whether that’s your company’s mission statement or a marriage proposal – then authenticity beats efficiency. Readers will want to know it’s come from you.
So do yourself a favour this Valentine’s Day and say it with flowers or chocolates if you must, but for goodness’ sake, say it in your own words.
If you’re trying to get the best out of AI in your teams, talk to us about our Intentional AI training.
Get in touch
Written by Hannah Moffatt, Creative Director at Definition
