When people speculate about a future in which we all give up on interpersonal relationships to shack up with robots I usually think, “Ah yes, another misanthrope who overestimates the capacity of machines to mimic the depth human feeling.” But today, I read an AI-generated love heart that said, “LOVE 2000 HOGS YEA,” and I’m reconsidering everything I thought I knew about romance.

“LOVE 2000 HOGS YEA” is just the start, too. If that doesn’t float your boat, how about the soon-to-be-timeless “HOW U HOT,” or the overwhelming power of “STANK LOVE” (which also happens to be an actual Outkast song with the lyrics “Stank love, stank love, stank love, stank love / Let me show your mind a new freaky side of love”). I mean, whether you prefer the straight-talking vibe of “YOU ARE BABE” or the inscrutable mystique of “LICK,” you have to admit this is one sweet talkin’ AI.

All of these phrases are the creation of a neural network prodded into life by Janelle Shane, a master of AI nonsense who’s previously worked her magic on Wikipedia titles, paint colors, and birds. As with these earlier examples, the neural network here was trained by imbibing a carefully constructed dataset. In this case, filled with the messages found on candy love hearts popular in the US and UK. The program looks for patterns in this data and then spits out what it thinks are plausible variations on the theme.

From here, you can tweak how far away the network roves from a sensible baseline. So, ask for a safe bet and you get messages like “LOVE BUN” and “CUTE KISS.” But as you change what’s known as the “temperature” of the network, these phrases get progressively stranger, as we swim from the shallows of plausibility and into the depths of the sublime. I’m talking “TIME HUG.” I’m talking “BOG LOVE,” and “TEAM BEAR.” I am talking, of course, “LOVE 2000 HOGS YEA.”

You can see a selection of AI-generated love hearts above, or check out more samples on Shane’s website. I warn you, though, after reading these messages you may never truly love a human again.