Unfinalizability is Bakhtin's ethical and ontological claim that the human being is never a completed fact, never fully knowable, never reducible to the sum of observable behaviors or definable characteristics. The living consciousness always retains a surplus, an excess, a dimension that escapes every attempt at total description. This is not a mystical claim but a phenomenological one: we experience ourselves and others as open, capable of surprise, able to become something we are not yet. To treat a person as finalized — as fully captured by a diagnosis, a category, a social role — is to commit a kind of violence, denying the open-ended character that makes the person human. In the AI context, unfinalizability becomes the ground of human dignity in an age of comprehensive categorization. The machine can describe, predict, and model human behavior with extraordinary precision; what it cannot do is acknowledge the dimension of the person that exceeds all models. The twelve-year-old's question — 'What am I for?' — is an assertion of unfinalizability, a refusal to be contained within any functional definition, an opening toward a future that has not been determined.
Bakhtin's clearest statement of unfinalizability appears in Problems of Dostoevsky's Poetics, where he argues that Dostoevsky's genius lay in granting his characters genuine freedom — allowing them to surprise the author, to develop in directions the plot did not predetermine, to escape every definition imposed on them. The Underground Man, Raskolnikov, Ivan Karamazov — each resists finalization, remains internally unresolved, capable of transformation up to the novel's final word. This is not a technical achievement but an ethical one: Dostoevsky respected his characters' unfinalizability because he respected the unfinalizability of actual human beings. The monologic author finalizes characters, turning them into objects that serve the author's predetermined meaning. The polyphonic author lets characters live.
In the AI age, unfinalizability faces its most sophisticated challenge. Machine learning systems excel at pattern recognition, prediction, and categorization — precisely the operations that treat persons as finalizable. The recommendation algorithm defines you by your past behavior. The resume-screening AI categorizes you by credentials. The performance-management system reduces you to metrics. Each acts as though the person were a completed fact whose future behavior can be extrapolated from past data. The error is not that the predictions are inaccurate (often they are accurate) but that they are finalizing — they treat the statistical tendency as the human truth, the pattern as the person. Bakhtin's framework insists: the person is always more than the pattern. The self that acts today can surprise the self that acted yesterday. Freedom is real, not as metaphysical premise but as lived reality.
The twelve-year-old's question documented in The Orange Pill — 'What am I for?' — is the unfinalizable human being asserting her openness against a culture that would reduce her to function. The question itself is the answer: you are for asking questions that exceed every answer, for becoming something you are not yet, for surprising the categories imposed on you. The machine can answer functional questions with precision (what skills are marketable, what careers are stable, what behaviors predict success). It cannot engage the existential dimension of the question, cannot acknowledge that the asking itself — the consciousness wondering about its own purpose — is the distinctively human contribution that no functional answer satisfies. This is not weakness but strength: the capacity to not-be-finalized, to remain open, to continue becoming, is the ground of human dignity in an age of intelligent machines.
The prescriptive implication: educational and organizational systems must resist the finalizing tendency of AI tools. The student is not the sum of her test scores; the worker is not the aggregate of her outputs; the person is not the pattern the algorithm detects. Systems that treat people as finalizable will systematically misunderstand, mismanage, and dehumanize those people. Systems that respect unfinalizability will design for surprise, will create space for development that metrics cannot capture, will maintain the irreducible gap between the person and every description of the person. This is not romanticization but realism: the unfinalizable person is the empirical reality; the finalized person is the administrative fiction.
Unfinalizability (nezavershennost', sometimes translated 'openness' or 'incompleteness') became central to Bakhtin's thought in the 1929 Dostoevsky book and remained a constant through his career. It connects to existentialist freedom (Sartre's 'existence precedes essence'), to phenomenological openness (Merleau-Ponty's embodied subject), and to the theological concept of the person as imago Dei — irreducible to any creaturely category.
The concept's relevance to AI became apparent in 2020s debates about algorithmic bias, predictive policing, and automated decision systems — all of which operate by treating persons as patterns rather than as open-ended beings capable of exceeding every pattern.
The person is never a completed fact. Consciousness retains a surplus that escapes every definition.
To finalize another is to deny their humanity. Treating a person as fully knowable commits ontological and ethical violence.
AI systems are structurally finalizing. Prediction, categorization, and optimization treat the person as a closed pattern.
The child's question asserts unfinalizability. 'What am I for?' refuses functional reduction.
Systems must design for openness. Respecting unfinalizability means creating space for the unpredictable, the developmental, the genuinely new.