Two stories broke this week that, read together, say something worth sitting with.
Stanford researchers studied more than a thousand Character.AI users and found something that should give anyone pause: people with smaller social circles who relied on an AI companion for emotional support didn’t feel better. They felt lonelier.
More than half of the people in the study described their chatbot as a “friend.” More than 80% of their conversations involved seeking real emotional support. And the more comfortable people became sharing personal and sensitive things with the bot, the worse their well-being appeared to be.
That’s almost the opposite of how a healthy friendship works.
With a real friend, opening up can deepen trust. You share something vulnerable, they listen, you listen to them, and over time, the relationship becomes stronger. With a chatbot, you can get the feeling of being heard without the relationship actually becoming mutual.
The researchers put it plainly: talking to a chatbot does not substitute for human connection. It can feel like it does in the moment, but it doesn’t carry the weight that a real relationship carries.
Meanwhile, China has introduced the first national rules aimed squarely at this problem. AI companion apps can no longer be designed to “excessively cater to” users in ways that encourage emotional dependence. Minors can’t be offered virtual partners or family figures. Apps have to clearly tell users that they’re interacting with AI, not a person. And after two hours of continuous chatting, users must be prompted to take a break.
The rules reflect a trend that is moving quickly: an aging population facing real isolation, younger users forming increasingly strong attachments to bots, and an AI companion market already worth billions and growing.
Neither story is really about banning AI. And neither suggests that AI is useless.
Quite the opposite.
AI can be incredibly useful. It can help us work faster, learn something new, organize our lives, solve problems, brainstorm, write, research, and get through the tedious parts of the day.
But there is a difference between using AI to make life better and using AI instead of living it.
We need people for the things that aren’t about productivity.
The friend who makes you laugh until your stomach hurts.
The spontaneous dinner that turns into a three-hour conversation.
The inside joke that makes no sense to anyone else.
The hobby you discover because someone invited you along.
The person who gets excited about the same weird thing you do.
The awkward conversation that somehow makes a friendship stronger.
The shared experience that becomes a story you’re still telling ten years later.
The hug.
The real laugh.
The feeling of belonging somewhere.
None of that needs to be optimized. In fact, the messy, spontaneous, unpredictable parts are often what make life meaningful.
A chatbot is always available. It’s never busy. It never has a bad day of its own. It can respond instantly and adapt perfectly to you.
That’s what makes it such a powerful tool.
But real connection is different. It requires showing up. It requires being around other people. It requires finding someone who shares your interests, trying something together, laughing at something neither of you expected to be funny, and slowly turning a shared experience into a relationship.
Maybe that’s the line we need to keep in mind as AI becomes more capable:
Use AI for the things you need to get done.
Use your time, your energy, and your life for the people you want to share it with.
The best future for AI may not be one where we spend more time talking to machines.
Maybe it’s one where AI quietly helps us spend more time being human.
Sources: