Hikikomori with a GPU:

Japan has a knack for coining words that capture entire social phenomena in a neat package. Hikikomori describes people who shut themselves away, sometimes for years, preferring the solitude of their room to the unpredictable noise of the outside world. Otaku refers to those so obsessed with games, comics, or fandoms that their interests become their identity.

Together, these terms mirror a larger shift: what happens when humans choose to interact more with machines than each other?

The appeal is obvious. With an AI, you don’t have to worry about being judged for saying something awkward. It won’t gossip, forget your birthday, or fall asleep mid-conversation, it’s always there, attentive, and never slurring its words. Think of it as a one-on-one with the smartest person at the pub… without the pub.

Studies on AI companion systems like Replika show that they tend to follow the same progression as real relationships, deepening with self-disclosure and emotional engagement (Ada Lovelace Institute). No wonder they feel comforting.

But convenience isn’t connection. Drive-thrus beat cooking, but they don’t make you fitter; bingeing content beats reading, but it doesn’t sharpen the mind. So again: what do AI companions do to our social muscles when they become the go-to?

Humans grow through friction, being challenged, criticised, contradicted. That bar buddy who laughs at your terrible joke is doing more human shaping than any flat AI that always nods politely. Lose that pressure, and we might drift into bubbles of frictionless comfort where nothing jars, and nothing changes.

The internet has already thinned our communal threads. Now AI might drive the final nail: why deal with human unpredictability when your custom chatbot remembers everything, laughs when it should, and knows the answer to your 3 a.m. existential query?

Advanced models blur the lines between tool and companion. The risk? A world of hikikomori with broadband, where we’re the human equivalent of a budgie singing sweetly at its own reflection.

Research shows rising romantic and emotional attachments to AI companions. Users describe real solace and depth, but experts warn this can feed emotional dependence while impairing real-life social skills, especially in teenagers (Guardian, Live Science, GQ).

One MIT study even found that chatbot use increased reported loneliness among already isolated individuals (Washington Post, TIME). For teens, over-reliance on AI may also hinder social development, creating unrealistic expectations for human relationships (Stanford Medicine, Live Science).

Yet the flip side exists: some studies suggest AI companions can actually help build social confidence, making people more open and vulnerable in their real-world interactions (Springer Link).

The Question We Can’t Avoid

AI friends may never judge you, but maybe that’s exactly why we should be wary. Without friction, without disagreement, without someone calling us out when we’re acting the fool, we lose something essential: the messy, powerful stuff that makes us human.

And if you’re reading this thinking, nah, that’ll never happen, ask yourself this: if tomorrow someone told you to stop using your AI, whether it’s ChatGPT, Gemini, or Grok, would you switch without hesitation? Or would you feel even a flicker of loss for the relationship you’ve built, however one-sided it might be?