At 8:10 every morning, Naoko became someone else.
She was a department manager when she entered the office: decisive, calm, economical with words. With her younger colleagues, she deliberately softened her voice. With executives, she became more formal. During video conferences, she kept the expression that colleagues had learned to interpret as confidence.
None of these versions felt entirely false.
That was what troubled her.
At home, she sometimes wondered which one was actually her.
The question had become harder in recent years because the workplace no longer saw only the person who arrived at the office. It saw a digital trail: messages, meeting transcripts, calendar patterns, performance evaluations, communication habits, and increasingly, AI-generated summaries of her work.
Her company had recently introduced an internal AI assistant that could prepare briefing notes in her preferred style. After several months, Naoko noticed something strange.
The system knew how she usually spoke.
It knew which phrases she favored, when she tended to disagree, which reports she considered important, and even how much detail she normally wanted before making a decision.
One afternoon, she asked it:
“Describe me.”
The answer appeared almost instantly.
“You are a cautious but pragmatic manager who values consensus while maintaining clear authority.”
Naoko stared at the screen.
It was accurate.
And yet she disliked it.
Not because it was wrong, but because it was incomplete in a way that felt dangerously precise.
It described the person she had become visible as.
It did not describe the woman who sometimes sat alone on the train after work and deliberately missed her stop because she did not want to go home yet.
It did not know about the books she bought but never finished, the decisions she regretted, or the irrational desire she occasionally had to abandon her carefully organized life and travel somewhere without telling anyone.
The machine had learned her social self.
But perhaps, Naoko thought, so had everyone else.
The sociologist Erving Goffman’s old metaphor of social life as a kind of performance suddenly seemed less metaphorical. A person moves between roles—manager, daughter, friend, customer, neighbor—and each setting carries expectations about how one should behave.
The surprising part was that modern technology had not eliminated this performance.
It had multiplied it.
A person could now have a workplace identity, a private social-media identity, a professional profile, an avatar, a collection of search histories, and potentially an AI system trained to imitate their patterns of speech and behavior. Recent work on digital identity argues that these representations are not simply detached masks: online and offline group memberships increasingly interact in forming how people understand and present themselves.
And the newest problem was that some of these representations could begin acting.
Researchers are now discussing “AI twins” and social-media clones: systems constructed from personal data that can reproduce aspects of a person’s knowledge, preferences, memories, language, or behavioral patterns. Their emergence raises a disturbing question—if a machine can convincingly reproduce the way you speak and make decisions, what exactly distinguishes the original from the representation?
Naoko tried an experiment.
She asked the company’s AI assistant to write an email refusing a request from a senior executive.
The result sounded exactly like her.
She deleted it.
Then she wrote the email herself.
It was almost identical.
She laughed.
Perhaps the machine had not stolen her personality.
Perhaps it had merely shown her something she had never wanted to admit:
She had become predictable.
That realization followed her home.
She stood before the bathroom mirror and looked at her own face. There was no mysterious inner person staring back at her. There was only a collection of memories, habits, relationships, bodily sensations, commitments, fears, ambitions, and stories about herself.
Which of them was the real Naoko?
The philosophical problem was ancient. The modern vocabulary was not.
Psychologists have long distinguished between aspects of the self-concept and the social identities attached to group membership and roles. Contemporary research is now asking how these processes change when digital platforms and AI become part of the social environment itself. AI systems can provide feedback, classification, and social signals that influence how people understand and regulate their identities. (Frontiers)
In other words, the mirror had acquired an algorithm.
And unlike an ordinary mirror, it remembered.
Naoko began to wonder whether the self had ever been a single thing.
Perhaps “I” was not the hidden object behind all the roles.
Perhaps “I” was the continuity connecting them.
The manager was real.
The daughter was real.
The exhausted woman on the train was real.
The carefully edited photograph was real in one sense, and artificial in another.
Even the digital version of herself might eventually become part of her identity—not because it was conscious or literally the same person, but because it carried information, relationships, expectations, and consequences that could reach back into her life. A recent philosophical analysis of “digital selves” makes precisely this possibility explicit: profiles, avatars, and AI replicas may be understood not merely as fictions, but as extensions or counterparts of a person’s continuing psychological identity. (Springer)
The question therefore changes.
“Who am I?” may be impossible to answer with a single description.
Perhaps a better question is:
“What continues to connect all the people I become?”
Naoko turned off the bathroom light.
Tomorrow morning, she would once again become the manager.
She would choose her words carefully. She would listen when she wanted to interrupt. She would smile when she was tired. She would perform the role because other people depended on her performing it.
But now she understood something she had not understood before.
The performance did not necessarily mean there was no self.
The self might be the thing that could move between performances without being completely contained by any of them.
And perhaps that was why the question “Who am I?” never disappeared.
A human being could answer it differently every morning and still remain, somehow, the same person.
The machine could reproduce her voice.
It could predict her decisions.
It might someday imitate her mannerisms so convincingly that other people could mistake it for her.
But there would still be one question the imitation could not settle for her:
Which of these versions does she choose to become?
That choice, Naoko thought, might be as close to a self as anyone ever gets.
All names of people and organizations appearing in this story are pseudonyms
Comments