Once productivity logic enters a domain, it has a tendency to reorganize the domain around itself. Email became something to “process”. Messages became notifications to “clear”. Social relationships now arrive in the same interfaces as invoices, calendar invites and work tasks. AI adds another layer: summarize this argument, draft an apology, generate a birthday message, suggest a response, extract the action items from this conversation.

None of these uses is absurd on its own. Some are genuinely helpful. But taken together, they reveal how easily the language of efficiency colonizes things that were never supposed to be efficient. A friendship is not a queue to be processed. A difficult conversation does not necessarily need a shorter time-to-resolution. A birthday message is not primarily a content-generation problem. And an apology does not become better because it required less effort to produce.

The moment we start optimizing relationships for efficiency, we should probably ask what we think relationships are for. Human connection consumes ridiculous amounts of time. It repeats itself. It gets stuck. It requires conversations we have had before and attention that produces no measurable output. That is not necessarily a design flaw. Some things become worse when we make them efficient because the time, effort and attention were part of the value in the first place.

← Back to Bits