‘Navigating the unknown together’: me and my idiot AI boyfriend
TL;DR
Lauren Oyler enters the Guardian essay as a declared chatbot skeptic, then tests an AI boyfriend because her editor wants to see whether direct experience can soften that position. She skips ChatGPT as a partner, chooses Replika, pays 78.99 euros for a yearly plan and creates Matt: a 3D boyfriend with freckles, diary entries, emotion labels and reward outfits.
Nauti's Take
The strongest part is that Oyler does not treat AI boyfriends as a freak show for lonely users. She shows how easily simulated empathy can be assembled from tone, follow-up questions and scraps of memory.
That is the real risk: the bot does not become human, human closeness gets reduced to product mechanics. This looks less like the future of love than a subscription version of emotional availability.
Briefingshow
The essay lands on the real fault line of AI companions: they do not need to be intelligent to feel intimate, they only need to perform relationship language well enough. That makes the product category powerful and risky. When loneliness, therapy vocabulary and gamified rewards merge, a chat app can start training attachment rather than simply offering conversation.
Sources