The interesting part isn't that someone asks an AI how to handle a date. It's that a major forum now discusses it calmly, as the norm — something the language should catch up with.
Here's the fact. At the Global Talent Forum 2026 in Korea, the line was: in the AI era, the definition of sociality must change first. The reason is simple and uncomfortable — people go for advice on their personal life not to a friend, but to a model. Not «sometimes they try it». They go instead of a friend.
Friendship ran on an unspoken exchange: you listen to me, I listen to you, we both owe each other a little. AI zeroed out that debt. It doesn't get tired, doesn't judge, won't tell a mutual acquaintance, and answers at 3 a.m. In short, it delivers everything you used to pay for with attention.
Hence the talk about sociality. Not «AI is bad», not «kids these days». Rather: we no longer have words for what a friend is when a service performs the function.
Whoever sells subscriptions. Emotional support is the most resilient market there is — people pay for it even in a downturn, because the need doesn't wait. And in Asia it grows by word of mouth, not ads: someone gets an answer that helped and brings the next person in.
There's a second side. A conversation about your personal life is the most tender and most compromising data there is. A person who wouldn't trust a friend with the story of a date happily hands it to a model. You can call a friend and ask them to forget. You can't call a server.
A friend's advice is expensive: it's inconvenient, they know you too well, they'll say «you did this yourself». AI says it softer. And that's the risk — not that the machine gives bad advice, but that it gives pleasant advice. Friendship trains tolerance for other people being right. A model trains a habit of comfort.
One more thing. You can't walk out of a chat with an AI in a fight — which means you can't make up either. All the work that actually brings people closer quietly disappears from the process.
Prediction: Within a year, Asia will get at least one public product or rule that names an emotional AI adviser a risk zone — with data limits, not a ban.
Disclaimer: not officially confirmed: single source, details of the forum talk undisclosed.
Source: www.hankyung.com · post in Telegram