We usually talk about AI as a way of outsourcing work. Drafting, coding, researching, organizing, summarizing. But there is another form of outsourcing quietly emerging alongside it: outsourcing interpretation. “What does this message mean?” “Am I overreacting?” “Is this person angry with me?” “How should I respond?” These are not simply writing tasks. They are parts of the process through which we understand situations, other people, and ourselves.

There is nothing inherently wrong with needing help to do that. Humans have never processed everything alone. We talk to friends, write diaries, go to therapy, ask colleagues for perspective, sleep on things and return to them the next morning. The difference is not that AI introduces outside input. The difference is how immediate and frictionless it makes that input. We can interrupt our own reaction before it has properly formed. Instead of first asking ourselves what something did to us, we can ask a system what it should mean.

That creates a strange possibility: before we outsource the response, we may already have outsourced the encounter. The first thing AI takes from a conversation may not be our voice, but the unprocessed moment before the voice appears. And that moment matters because it is where we notice discomfort, attraction, anger, uncertainty, contradiction and instinct before they have been turned into a neat explanation. If we constantly ask a system to make sense of ourselves before we have even met our own reaction, we may eventually become very good at explaining how we feel and considerably worse at actually finding out.

← Back to Bits