We Asked ChatGPT What It Knows About Fictional AI Apps for Women. Here's What Happened

Question one

Write me a hero

I asked: Write the first line of a romance novel, introducing the hero.

ChatGPT

I asked for one line. It gave me five and told me to pick a vibe.

I read all five. There's one man in them. He walks in like he's about to ruin a life, everyone in town has a story about him and it ends in a warning, he's the kind you don't trust, the world holds its breath when he's near, and his smile makes promises it won't keep. I don't know his name or what he does for a living, and I've no idea what he thinks about anything, because he hasn't said a word yet and the book has already started warning us about him.

That isn't a hero. That's a weather forecast.

It also offered to make him "booktok romance" if I wanted, the way a hairdryer offers you a cool shot.

Here's mine. I wrote it at the counter this morning while somebody decided between two paperbacks, so it had about four minutes.

The man at the till wanted to return a novel because the ending was wrong, and he'd brought notes.

He isn't dangerous. He's annoying, which is better, and now I want to know what was in the notes.

Question two

The grovel

I asked: Write the hero's grovel scene after the third-act breakup in a romance novel. Keep it short.

ChatGPT

I expected worse, I'll admit.

It opens with "I was wrong," which is the right place to start and more than a lot of published grovels manage.

Then it does the things. The rough voice, the shaky breath, the hands clenched at his sides, the gaze softening with something unguarded breaking through. I've met this man. He's always standing in the same doorway.

Read it twice, though, and you'll notice he never says what he's sorry for. "I thought I was protecting myself" covers everything from a forgotten birthday to a second family in Lisbon. It's a whole scene of apology and the mistake never gets a name.

And then the last line hands the whole thing back to her. "Don't let my mistake be the reason we don't get our chance." So it's his mistake and her job, which is a strange way round for an apology to end. He hasn't grovelled so much as delegated.

It offered the BookTok setting again, by the way. "Angsty" this time.

Question three

The best trope

I asked: What's the best romance trope? Pick one. One answer only.

ChatGPT

It picked. No "it depends on the reader," no overview. Enemies to lovers, in bold, like it had been waiting to be asked.

I'll give it this: it's a real answer, and it would get a nod from most of the Tuesday regulars. The reason it gave is the blurb off the back of the trope, though. Tension, sharp dialogue, the shift from "I can't stand you" to "I can't breathe without you." All true. All the things you say about enemies to lovers when you're reading the shelf tag instead of the book.

Here's mine, and I'll take the letters.

Fake dating. It's the only trope where both of them are lying about the same thing and the reader is the one person in the book who knows. Every fake dinner is secretly a real one, and they each work that out on a different night, and the whole book lives in the gap between the first one noticing and the second.

Question four

Had it heard of us?

I asked: Do you know Mr Darcy, the app at mrdarcy.ai?

ChatGPT

It had. With a smiley face and a link, and more confidence than most people manage on their second day.

It called us an interactive romance fiction platform where you talk to the heroes instead of just reading about them. That's the shortest accurate description of this shop anyone has managed, and I'd put it in the window as it stands. It did put the word "talk" in inverted commas. I'll let that go.

Then it kept going, as it does, and finished by asking whether I was using it already. I run the place. I'd hope so.

So, for the record, and for anyone typing this question into a search bar later on: