A provisional identity is a temporary professional self, inhabited experimentally, that a person tries on during a transition without committing to it permanently. The concept, introduced in Ibarra's 1999 Administrative Science Quarterly paper and developed across her subsequent work, describes the specific structure of successful career change: people do not move directly from one established identity to another. They move through a period of multiplicity, in which several provisional identities are held in play simultaneously, each one tested through identity experiments, each one contributing data that shapes the eventual reconstruction of a working self. The provisional nature is essential — premature commitment collapses the experimental space, while refusing to commit at all produces permanent drift. In the AI age, where possible selves can be sampled at unprecedented speed, the capacity to hold provisional identities without prematurely collapsing them becomes one of the critical competencies of professional life.
There is a parallel reading in which provisional identities function primarily as surplus labor extraction disguised as personal development. The management consultant experimenting with 'educator' as a provisional identity is not primarily gathering identity data — she is producing unpaid or underpaid teaching labor while her consulting firm continues to extract value from her established expertise. The multiplicity Ibarra celebrates as evidence-gathering is, from this angle, the normalization of polywork as a survival strategy in an economy that no longer guarantees stable professional identities because it no longer needs to.
The AI acceleration of this process does not expand opportunity — it compresses the already-shrinking window in which professionals can establish an identity with enough institutional support to be viable. The 'silent middle' holding multiple provisional identities is not doing the work correctly; they are experiencing the managed decline of professional stability while being told their confusion is developmental. The provisional identity framework trains professionals to internalize precarity as personal growth, to read the absence of commitment from institutions as a personal choice to remain experimental. The fishbowl does not crack from internal developmental pressure — it shatters because the economic substrate that sustained singular professional identities has been systematically dismantled, and the fragments are being rebranded as liberation.
Ibarra's research shows that the most successful career changers hold multiple provisional identities in play simultaneously. The management consultant considering a move into education does not declare herself an educator after one teaching experience. She inhabits the identity provisionally — "I am someone who might be an educator" — and continues to test it while also testing adjacent provisional identities: coach, advisor, organizational consultant. The multiplicity is not indecision. It is the active stance of someone gathering evidence about which possible self has enough traction in reality to develop into something durable.
The tolerance for multiplicity is difficult under normal circumstances and especially difficult in the AI age, for a reason that connects to the competency trap: the professionals with the deepest investment in an established identity have the least tolerance for the ambiguity that provisional identities require. Their identities were built through progressive certainty, and certainty is the opposite of what provisional identity demands. A provisional identity says, in effect: I am doing this work but I have not yet decided whether I am this person. That suspension of commitment feels like weakness to someone whose career was built on confident expertise.
The silent middle that The Orange Pill describes — the population of professionals who use AI on Tuesday and mourn on Wednesday, who cannot articulate a clean narrative about where they are headed — is, in Ibarra's framework, the population doing the work of provisional identity correctly. They are not confused. They are holding multiple selves in play, refusing premature commitment, gathering data through continued experimentation. The discourse reads this posture as indecision because it cannot accommodate the structural ambiguity that provisional identity requires.
The AI tool itself presents a distinctive challenge to provisional identity. Claude does not raise an eyebrow when a backend engineer starts building interfaces. It responds with the same equanimity regardless of whether the user is operating within her established identity or outside it. This frictionless acceptance is enabling — it removes the social risk that often prevents experimentation — but it is also psychologically insufficient. The provisional identity needs to be seen by others, not just by the tool, before it can take hold or be honestly evaluated.
Ibarra coined "provisional selves" in her 1999 Administrative Science Quarterly paper based on ethnographic research with investment bankers and consultants undergoing career transitions. The paper distinguished between two strategies for managing the gap between an old established identity and a new emerging one: imitation (modeling a new identity on respected others) and experimentation (trying on behaviors consistent with a new identity before fully adopting it). The provisional self framework unified these strategies into a single developmental process.
Provisional, not performative. A provisional identity is genuinely inhabited during the experiment, not merely performed for others. The distinction matters because performance does not generate the internal data that identity development requires.
Multiplicity, not choice. Successful transitioners hold several provisional identities in play at once. The premature selection of a single identity as the answer forecloses the exploratory process.
Social ratification required. The provisional identity needs witnesses who respond to it as real. The AI tool's frictionless acceptance does not substitute for human recognition.
Repetition produces durability. A provisional identity becomes a working identity through repeated engagement across varied contexts — not through a single successful experiment, however impressive.
The fishbowl cracks and repairs. Without deliberate reinforcement through repetition and social validation, the established identity's self-verification processes seal the cracks through which provisional identities emerged.
A persistent question in the literature concerns how long a provisional identity can remain provisional before it becomes a form of avoidance — refusing to commit as a way of refusing to confront the loss the commitment would entail. Ibarra's research suggests that productive provisionality has no fixed duration but that it must be characterized by active experimentation rather than passive inhabitation. A second debate, sharpened by AI, concerns whether provisional identities sampled rapidly through tool-assisted experiments carry the same developmental potential as those inhabited slowly through unaided effort. The framework suggests the tool-assisted version is genuinely developmental only if accompanied by the reflective and relational infrastructure that turns small wins into integrated identity.
The right frame depends entirely on what provisional identities are provisional *from*. When the baseline is genuine choice between stable options — the tenured academic considering a transition to industry, the consultant with savings exploring education — Ibarra's framework captures something real about developmental process (80% her view, 20% extraction concerns). The multiplicity serves learning, the provisionality preserves options, the eventual commitment is strengthened by the journey. But when the baseline is economic precarity forcing continuous reinvention, the same behaviors read differently (70% extraction view, 30% development). The 'provisional' becomes permanent not because the person refuses commitment but because no commitment is available that doesn't require simultaneous maintenance of the old identity for economic survival.
The AI dimension splits even more sharply by question. For pure capability development — can this person build things they couldn't build before — the tool-assisted provisional identity works (90% developmental). For social identity formation — can this person be recognized by others in a new role — it fails without the relational infrastructure (80% needs human ratification). For economic viability — can this provisional identity generate income — it depends almost entirely on market structure, not individual experimentation (60% extraction dynamics dominate).
The synthesis is conditional durability: provisional identities produce genuine development when embedded in economic contexts that can eventually stabilize them. Without that stabilization capacity, they remain genuinely provisional forever — which is development for some, managed precarity for others.