"Well, cuts, already taken?" That is the level on which tear-open slogans usually move: at most suitable as a title for an anthology about a bad dating experience, but hardly for really getting into conversation with someone.

But what happens when machines, rather than people, “think up” things?

Can you even do that?

Will the sayings then be different, maybe even better?

Or do they sound even stiffer, more clumsy (if that's even possible)?

The scientist Janelle Shane, who has published an explanatory book on Artificial Intelligence (AI) and regularly writes on the subject for various media, simply tried it out.

On her humorous blog aiweirdness.com, she describes how, out of joke, she commissioned the text generator GPT-3 to create “top pick-up lines” in order to “impress his crush”.

GPT-3 is a so-called language model that can update texts in such a way that they read almost as if they had been written by a person.

The special thing about GPT-3 and what makes its texts appear at least as similar to those written by humans: He can be funny and quick-witted.

Two talents that can also be extremely helpful when flirting.

But is that enough?

Could the computer-generated sayings be typed into any dating app without hesitation if one couldn't think of anything?

Or say it in a bar as soon as it reopens one day?


The short answer is no.

The longer one: It all depends.

Because Shane has tested four differently powerful variants of GPT-3 and accordingly the quality of the results varies greatly.

Ada, for example, the smallest variant according to Shane, only produced incoherent word combinations.

“Embroidery tags”, “MONTINA”, “Body Softening Pads” - even with a lot of imagination you can't read out an attempt at flirting here.

With the other three variants, which are called Babbage, Curie and DaVinci and are more competent than Ada, things look different.

They mostly generated whole sentences that make sense.

Some follow the pattern of a compliment, others could be imagined as starting a conversation ("Do you like ... pancakes?").

What all sayings have in common is that they create surprising connections.

It's about dogs in trench coats, beautiful fangs or air fresheners - all things that are not immediately associated with (sexual) attraction and love.

But in the words of Babbage, Curie and DaVinci, they develop an idiosyncratic poetry that sometimes makes you think, but is often simply funny (“It is urgent that you become a professional athlete”).

Poetry or joke, you still have to be a bit crazy to actually use the AI's pick-up lines when flirting?

That's right.

But better than “Well, are you already occupied?”, For example, “You look good today.

Do you want snacks? ”In any case.

DaVinci's best sayings


I love you.

I don't care that you are a dog in a trench coat.

You have the most beautiful fangs I've ever seen.

Do you know what i like about you

Your ... long ... legs ...


Do you like ... pancakes?

I have exactly four stickers.

I want you to be the fifth.

You have a wonderful face.

Can i put it on an air freshener?

I want your smell always close to me.

I will briefly summarize the plot of "Back to the Future II" for you.

The best sayings from Curie

Your eyes are like two rainbows and a rainbow of eyes.

I can't help but stare at her.

I'm like ice cream ... you can keep me in the fridge for a while, but then I'll melt!

You have the best french toast I've ever had!


I picked some nice flowers.

Do you want to smell it?

Here, try to release my hand from them.

Hey, my name is john smith.

Will you sit on my bread bin while I cook or is there a speed limit for this thing?

My name is a complicated combination of 45 degrees of forward movement, 25 degrees of deviation to the left, and 75 degrees of acceleration upwards and infinity, and that's where my love for you stops.

Babbage's best sayings

You look good today.

Do you want snacks?

It is imperative that you become a professional athlete.

Did you steal anything today?

Will you marry me?

Here you will find the sayings in the original English.