AI Is Not Replacing People – It's Replacing Availability

Whenever I wanted a second opinion on something, I used to go to one or two close friends, or to my partner. Nowadays, I find myself more and more consulting AI assistants like Claude or ChatGPT. I know they aren't human, don't have feelings, and don't have the shared history with me like my friends and partner do. However, for someone like me, who hems and haws over whether I SHOULD reach out to someone, and who gets drained by social situations, using AI is an attractive option that eliminates social friction.

We have limited bandwidth to spend on our own lives and obligations. We get tired, busy, distracted, and overwhelmed. Even when my connection to someone is strong, I still go through a decision tree in my head: When is it appropriate to reach out? Should I wait for a better time? Should I flesh out this idea on my own more thoroughly first? This mental calculation itself can be exhausting.

A fully available friend still costs something to ask. AI assistants like ChatGPT and Claude are never busy, and won't keep a mental scorecard. I don’t have to wonder whether now is a good time, or whether I'm imposing. Eliminating this "social transaction cost" can be compelling on its own as a reason to lean in more to using AI assistants.

There are drawbacks, though. When we can externalize a thought, a question, a draft, or a half-formed idea without worrying about the cost to another person, there is less incentive to let the idea marinate, to chew on it, to learn how to explain it to someone else to get genuine feedback. We can simply offload every stray thought and idea to a system designed to respond. On the one hand, this is a pretty cool thing, having the ability to flesh out any idea quickly, which can really speed up thinking and allow us to explore a lot of different interests. On the other hand, over time, it can create an isolating effect, where there is less reason to interact with other human beings, instead choosing to use the always-available AI.

This isn't a new phenomenon. Internet forums, instant messaging, texting, and social media each amplified the ability to get information and to communicate more quickly as well as asynchronously. AI takes this ability even further through immediate responsiveness, bolstered through access to the Internet and memories of the user's interests. It doesn’t merely host words on a page to be viewed by other human beings later. It directly answers them without human involvement, and in a way that feels conversational.

Human connection naturally has bumps, with occasional misunderstandings and arguments, even heated disagreements. This is part of the social friction that I anticipate when deciding whether to reach out to someone. There's something comforting and reassuring in having an AI assistant in my pocket who, by design, is always on my side. AI feels safe in that sense; there are no relational stakes involved in choosing to spitball an idea at Claude. But sometimes we need to hear hard truths to grow, and AI assistants can't, and shouldn't, replace that human ability.

Rather than asking whether we should use AI, because it's not going anywhere... it may be more useful to ask something else. Where are we willing to allow AI into our lives?

How and when we decide to use AI will differ person to person, but we must be intentional with our AI usage. Otherwise, we may wake up one day and realize we haven't talked to a human in weeks. And we may even be OK with that.