IN THE MIRROR: REWRITING OF THE SELF. THE INTERACTION BETWEEN AUTOBIOGRAPHICAL PRATICE AND GENERATIVE ARTIFICIAL INTELLIGENCE.

Abstract

The spread of generative artificial intelligence systems opens new questions also within the field of autobiographical practices. Writing about oneself, traditionally understood as a space for reflection, distancing, and attribution of meaning to lived experience, today encounters tools capable of intervening on the text, modifying its tone, amplifying some of its aspects and proposing different narrative perspectives. The possibility of receiving a reworked version of one’s own account introduces a further element into the autobiographical process: the subject no longer confronts only what he or she has written, but also a variant of his or her own text that appears to him or her, at the same time, familiar and foreign.

The present contribution examines this space of tension. Attention is directed rather to the reactions that such rewriting may arouse, algorithmic rewriting therefore appears as a mediating device. The present work is based on the hypothesis that algorithmic rewriting may have a pedagogical value if understood as a mediating device that may activate new questions in subjects about their own account.

The narrative identity of each person is not immutable, but is constructed through interpretation and autobiographical reflection (Ricoeur, 1988).

With the introduction of chatbots based on Large Language Models, technology produces linguistic responses becoming a “quasi-other” (Osler, 2026), an artificial interlocutor capable of generating real effects on conversation. Communication can therefore produce effects even in the absence of an intentionality assimilable to the human one (Rivoltella, 2024). The algorithmic rewriting of autobiographical texts has no value in itself, its value depends on the relationship that the subject establishes with the returned text.

This is the “hermeneutic paradox”: ChatGPT’s algorithmic rewriting can produce reflective effects even when the user knows that they do not derive from an authentic understanding of his or her experience, the chatbot’s response can be useful not because it is true, but because it induces the person to look at his or her own history from another perspective and to formulate new questions (Kvietkute & Ness, 2026).

The risk is precisely that of delegating to the machine the authority to tell who we are (Osler, 2026). In the present work, AI is therefore considered as a symbolic device of mediation. The machine returns to the subject a text that is, at the same time, recognizable and altered and the underlying hypothesis is that such a gap may activate reflective effects.

The present study is qualitative and exploratory and has the objective of observing in a circumscribed way the reactions that emerge when a subject encounters an algorithmically reworked version of his or her own autobiographical text.

For each text, three variants were produced, corresponding to three prompting protocols: emotional intensification; narrative distancing; change of point of view. The corpus analyzed includes: 3 original autobiographical texts; 9 rewriting generated by the system; 3 qualitative questionnaires completed by the participants.

The feedback forms were analyzed through a qualitative content analysis with a deductive orientation (Hsieh and Shannon, 2005). In the construction of the feedback forms, four dimensions were chosen that oriented the participants’ reflection: relationship between recognition and estrangement; explicitation of implicit meanings; effects of distancing and change of perspective; attribution of authenticity and interpretive authority to algorithmic rewriting.

The encounter with the algorithmic rewriting produces in the three cases a tension between familiarity and estrangement. The bodily reactions show that reading is not experienced as a simple exercise in textual comparison, but gives rise to sensations of emotion and surprise. Algorithmic rewriting is perceived as a text that is one’s own and not one’s own, its reflective capacity resides precisely in this gap.

The participants report that the algorithmic rewriting broaden the perception of the lived experience, but do not transform autobiographical memory, nor add forgotten elements to the memory. In the responses of the first two participants, the rewriting in the third person is perceived as cold and incapable of returning the emotional density of the experience, while in those of the third participant the distancing allowed recognition, an effect of validation.

The attitude of the participants is devoid of “deference” toward the machine; on the contrary, they receive the rewriting critically, always comparing them with their own lived experience and keeping attention high on passages perceived as inaccurate, artificial, or excessively interpretive.

Although the study presents evident limitations linked to the small number of participants and to its exploratory function, it can be stated that, in the three case studies, algorithmic rewriting functioned as a reflective device. Its reflective value does not reside so much in the capacity to reformulate more or less authentic readings of autobiographical experience, as in the reaction produced in the subject.

Algorithmic rewriting can be interpreted as a form of narrative mirror: it does not necessarily return a truth about the subject, but produces a textual configuration that the subject is called upon to recognize, contest, or reject. Precisely for this reason, the pedagogical value of AI should not be sought in the authority of the output, but in the possibility of activating critical work on the distance between the original account and its artificial reworking (Vallor, 2024; Fabry, 2025).

AI can be used as a tool of mediation, capable of producing narrative variants and generating questions, but it is the subject who must carry out the hermeneutic work on his or her own experience.