The message on Niko’s phone is courteous, clear, and almost impossible to argue with. It thanks a friend for the years they have known each other, explains that he needs some distance, and wishes her well. An AI assistant wrote it after he typed, “How do I tell her I can’t keep being the person she calls at two in the morning?” He reads the reply three times. Every sentence is sensible. None sounds like him.
I’m Camilla Wellton, and I want to look at Niko’s message with you, honey. I founded the Swedish Institute of Inner Technology because I care about the capacities we carry into the age of AI. But forget the conference stage for a minute. It is late, your friend needs help, and you need sleep. A machine has produced an immaculate message for Niko. He reads it and feels himself vanish from the sentence. Have you felt that? Sometimes the words are courteous, clear and completely wrong for the person who must live with them. Let’s find out what the draft can do for him and what only he can decide.
Maybe the polished message is too final. Niko does not want to end the friendship. He wants to stop waking in panic when the phone lights up. The draft has quietly translated I cannot keep doing this into I am leaving. That is a plausible translation, not an inevitable one. A tool can be helpful and still make a choice the person using it has not chosen.
What would you write if you were Niko, knowing that your friend will hear your voice in the message? He writes a rougher sentence himself: “I care about you. I can talk tomorrow, but I can’t keep answering these calls in the middle of the night.” He hesitates before sending it. Perhaps his friend is in immediate danger. If so, she needs urgent human support, and he needs to help her reach it rather than become a private emergency service on his own. Perhaps she is lonely and afraid, and the calls have become a habit. The response will be different depending on what is actually happening. The screen cannot settle that for him.
This is what I mean when I ask what a tool cannot want. It can predict a useful phrase from many examples. It cannot be Niko’s friend, inhabit his fatigue, remember a promise he made last winter, or bear the consequence of a sentence sent tonight. It may help him see options. The desire, judgment and responsibility remain in a human relationship.
My book and card deck Reality Revealed ask us to examine the conclusions that have begun to feel like reality. Niko’s old conclusion is “If I don’t answer, I am abandoning her.” There may be a history behind it. Perhaps someone once needed help and he missed the call. Perhaps his family praised him for being available when no one else was. He can take that history seriously without asking it to make every decision in the present. What support does his friend need now? What can he actually give? Who else can be involved? These questions are less fluent than the draft, and much more useful.
If you have used a tool to ask whether you are selfish, you may recognize a second story on the phone. Niko has asked the tool whether he is selfish. It answers reassuringly. He asks again with a different prompt, hoping for more certainty. A machine that responds every time can become an excellent companion for a question you are afraid to bring to another person. It can also keep the question circling. Niko might need to talk with his friend, or a trusted person who knows both of them, before another paragraph of reassurance will mean anything.
I do use tools. A translation may help two people find a common starting point. A draft can reveal that what I thought was clear is confusing. Someone who struggles to find words may be able to say more with a little assistance. The problem begins when ease is mistaken for authorship of the choice. A sentence can be improved until it no longer carries the truth the writer needed to say.
In Seduce My Mind, one of my Sensual Hero’s Journey books, I ask how sensation becomes interpretation, and interpretation becomes a story. Niko feels a jolt when the phone rings. “She needs me,” he thinks. Then: “If I don’t answer, something terrible will happen.” A tool can help him list other interpretations. He may still have to notice the jolt in his body, check what he knows about this particular night, and choose an answer that is neither automatic rescue nor cold withdrawal.
The distinction matters outside friendship. A designer can ask a model for thirty names and still have to hear which one belongs to the work. A parent can ask for a letter to a school and still decide whether it speaks respectfully about their child. A reader can receive an elegant summary of a book and still miss the sentence that would have changed their mind if they had met it themselves. What we outsource is not only effort. Sometimes it is the very encounter in which we learn what we think.
That does not mean every task must become an ordeal of authenticity. Let a machine sort the calendar. Let it offer a first draft when the blank page is a wall. Then return to the point where your judgment is needed. Ask what the output assumes, whose interests it serves, and whether you would still stand by the result when the person affected answers back.
You may decide to use the draft as a rehearsal rather than a message. Niko saves it but does not send it. He writes to his friend in his own uneven language. She responds later, hurt at first, then curious about what he can offer. They arrange a time to talk. They also name another person she can call when the night becomes difficult. Nothing is fixed by a single message. Niko has said something true enough for the friendship to work with it.
If she had answered differently, the message would still have mattered. Human agency does not guarantee a pleasing reply. It means Niko can recognize the words as his and remain present for the conversation they begin. That is a capacity worth practicing while assistance becomes easier to obtain.
