Living with an AI Twin: what it changes about identity

The first week with an AI Twin is a novelty. You ask it things you already know the answer to, to see whether it sounds like you. It mostly does, and that is briefly amazing and then no longer interesting.

The months after that are where the real questions live, and none of them are technical.

Someone else is now speaking as you

The moment a public link exists, people you have never met are having conversations with something wearing your name. They are not confused about it: they know they are talking to an AI Twin. They still come away with an impression of you.

This is not new in kind. A book does the same thing. A recorded talk does the same thing. What is new is that the representation is responsive. It answers the specific question the person actually had, which means it goes places no book ever went, and it does so without you present to steer it.

The practical consequence is that you become responsible for a range of possible answers rather than a fixed set of published ones. That is a different editorial job.

You find out what you actually think

The most useful side effect is unflattering. Reading your AI Twin's answers to real questions surfaces every place where your material is vague, contradictory, or ten years out of date.

You wrote something in 2019 that you no longer believe. You never retracted it, because nobody was reading it closely. Your AI Twin reads it closely, and now it tells strangers about it.

Curating an AI Twin turns into curating your own body of work. Most people have never done that deliberately.

Where the boundary has to sit

The rule that has held up: an AI Twin should represent what you know, never simulate what you would decide.

Knowledge scales. Judgement does not. Judgement depends on context the AI Twin does not have: who is asking, what they are not saying, what happened last time, what you owe the person. An AI Twin that guesses at judgement will eventually be confidently wrong in a way that costs you something real.

So the design goal is a system that knows the edges of its own knowledge and hands the rest back. When it says I do not have that, ask me directly, that is not a failure. That is the feature working.

What it changes about your time

The honest accounting is not that an AI Twin saves you hours in the abstract. It is that it removes a specific category of work: the explanation you have given four hundred times.

What arrives instead is a smaller number of better conversations. The people who reach you have already had the introductory conversation with your AI Twin. They arrive knowing what you do, which means the first ten minutes are no longer spent establishing it.

The identity question, answered practically

People ask whether an AI Twin is you. It is not, and pretending otherwise is where these products go wrong.

It is a representation you authored, the way a CV is a representation you authored, except it answers back. You are responsible for what it says because you chose the material. That responsibility is the whole reason ownership of the material matters: you cannot be accountable for a representation you do not control.

This piece revisits an earlier essay from our Medium archive: Living with my AI Twin.

What an AI Twin actually is and how ownership works.