Identity Experiments

In the AI Story

The traditional identity experiment was small, deliberate, and slow. A management consultant considering the nonprofit sector did not quit McKinsey and apply to the Red Cross. She volunteered on weekends, attended conferences, took pro bono engagements. Each foray was a data point. Does this feel right? Am I energized by this, or merely attracted to the idea of being energized by it? The slowness was functionally necessary — each experiment generated a complex signal that included cognitive, emotional, and social elements, and processing the signal took time.

Ibarra's research shows that the most valuable reflection often happens between experiments, in the quiet intervals when the person is not trying anything new but is digesting what the last experiment revealed. The interval between experiments is where identity data gets integrated into the working self. AI has collapsed this interval. The designer at Napster who, within two weeks of working with Claude Code, was building complete features end to end ran what was functionally an identity experiment in days rather than months.

The collapsed timeline produces what might be called a data-integration mismatch. Experiments generate identity data faster than the human identity system can absorb it. Each experiment produces a signal; the signals accumulate; but the integration mechanism — the slow, reflective, often unconscious process by which a person incorporates new experiences into a coherent sense of self — cannot accelerate to match the pace of input. The result is a buffer overflow: more identity data than the system can process, more possible selves in play than reflective capacity can evaluate.

Ibarra's prescription is not to reduce the speed of experimentation — the abundance of possible selves is genuinely valuable and democratization of access to them is genuinely important. The prescription is to build structures that support integration alongside experimentation: structured reflection after experiments, deliberate return to possible selves that showed early promise, conversations with trusted others about what the experiments revealed. These structures do not arise naturally in an environment optimized for speed.

The distinction between identity tourism and identity development turns on whether experiments are repeated across varied conditions until a provisional identity develops durable weight, or whether each experiment is abandoned for the next exciting possibility. The AI tool's frictionless capacity to support any experiment makes tourism the default trajectory unless the practitioner deliberately invests in the return — the discipline of going back to the same provisional identity rather than moving on.

Origin

Ibarra developed the identity experiments framework across Working Identity (2003) and subsequent publications, drawing on ethnographic case studies of professionals in transition. The approach unified earlier research on imitation and experimentation as strategies for provisional self-construction with organizational scholarship on learning-by-doing and experiential knowledge acquisition.

Key Ideas

Concrete and embodied, not imagined. The experiment requires actual action in a real context that produces real consequences, however small-scale.

Reversible. The experiment does not demand commitment; it is designed to generate data about whether commitment would be warranted.

Generates multi-dimensional signal. Each experiment produces cognitive data (Can I do this?), emotional data (Does this feel right?), and social data (Do these feel like my people?). Processing all three dimensions takes time.

Integration happens between experiments. The interval, not the experiment itself, is where identity development consolidates. AI has shrunk the interval in ways that threaten consolidation.

Repetition across varied conditions. A single successful experiment is a data point. A provisional identity becomes a working identity only through repeated engagement — particularly through experiments that include difficulty, tedium, and failure, not only excitement.

Debates & Critiques

A central debate concerns whether AI-mediated identity experiments can produce identity change equivalent to unmediated ones. The skeptic argues that an experiment in which the tool does most of the cognitive work — producing the code, drafting the analysis, generating the output — tests capability but not identity, because identity is forged through the specific struggle that the tool eliminates. The proponent counters that the identity-forming experience is the direction of the work, not the execution, and that AI-mediated experiments expose the practitioner to the evaluative, judgment, and integrative experiences that constitute the new identity in the judgment economy. Both positions contain truth; the evidence suggests the question resolves differently for different kinds of transitions.