Impostor syndrome and AI: when generative AI makes you feel like a fraud

Generative AI activates and intensifies impostor phenomenon — the persistent sense of being less competent than you appear — through a specific mechanism. When tools like ChatGPT, Claude and Gemini produce outputs comparable to years of professional skill, they appear to confirm what impostor phenomenon creates: that the professional's competence was never real. Research by Wafaa Albadry (IADT Dublin, 2026), published in Springer Nature's Governing Artificial Intelligence: Power, Legitimacy, and Democratic Futures, measured this relationship directly using the Clance Impostor Phenomenon Scale in creative and knowledge professionals. It is not a personal failing. It is a predictable psychological response to a specific professional disruption — and it is addressable.

What impostor phenomenon is

Impostor phenomenon — the preferred clinical term, chosen to avoid the pathologising implication of "syndrome" — is the persistent internal experience of intellectual fraudulence that persists despite external evidence of competence. It was first described by psychologists Pauline Clance and Suzanne Imes (1978) in a study of high-achieving women in academic settings. Subsequent research established that it affects approximately 70% of people at some point in their working lives. It is significantly more persistent and has greater professional consequences for women and for professionals from underrepresented groups.

The core experience has three components: the belief that success is due to luck or external factors rather than genuine skill; fear that real competence will be required and the inability to demonstrate it will expose the fraud; and a pattern of dismissing success while treating failure as confirming evidence. It is measured using the Clance Impostor Phenomenon Scale (CIPS), a validated 20-item self-report instrument.

How generative AI triggers it

Before 2022, impostor phenomenon operated in the gap between perceived external standards and perceived internal capability. The professional feared being exposed as less competent than peers. Generative AI introduces a new reference point: a tool that produces, without effort or training, the exact outputs the professional has spent years mastering. The threat is no longer "other people are more competent than me." It is "a tool anyone can access is producing what I do."

For someone experiencing impostor phenomenon, this functions as apparent confirmation: "It was never real. Your skills were a slower, more expensive version of this." For someone without pre-existing impostor phenomenon, the same experience can create a new form of it: "If the tool can do this, what exactly is my professional value?"

Cyberpsychology frames this as an identity threat response: a situation in which a core component of professional self-concept — "I am someone who writes well, thinks clearly, communicates effectively" — is perceived as no longer uniquely human. The response activates the same threat-appraisal mechanisms as any perceived attack on identity.

The research

Wafaa Albadry's MSc research at IADT Dublin (2026), published in Springer Nature's Governing Artificial Intelligence: Power, Legitimacy, and Democratic Futures, is among the first empirical studies to measure the relationship between generative AI adoption and impostor phenomenon directly. Using the Clance Impostor Phenomenon Scale and validated AI anxiety measures, the research found: a significant positive correlation between generative AI tool adoption and impostor phenomenon scores in knowledge workers and creative professionals; an amplification effect among women in creative industries following intensive AI tool exposure; and a protective effect of psychological safety interventions — participants who received structured support for the psychological dimensions of AI adoption showed significantly lower anxiety responses than those who received practical tool training alone.

Why this disruption is different from previous automation

Automation has disrupted professional practice repeatedly. What makes generative AI different — and why it triggers impostor phenomenon responses when previous waves often did not — is its domain: professional identity work.

Previous automation replaced physical and procedural tasks. Professionals did these tasks but were not defined by them. Generative AI targets the outputs of creative and knowledge work: writing, analysis, research synthesis, translation, communication design. These are the tasks professional identity is built on. "I am a journalist." "I am a researcher." "I am a communicator." When AI produces a journalist's output, the threat is to the journalist's identity, not only their workflow. This is not sector-specific: HR professionals, lawyers, marketers, researchers, educators and consultants face the same displacement of identity-central work.

The disclosure loop impostor phenomenon creates

In AI contexts, impostor phenomenon creates a specific damaging loop around disclosure. Professionals who fear that AI use signals inability to do their own work face two options: disclose AI use and risk being perceived as compensating for inadequate skill — confirming the impostor fear; or don't disclose, use AI anyway, and live with the concealment that intensifies impostor phenomenon. Neither resolves the underlying experience. The resolution is an organisational culture that separates AI disclosure from competence judgement entirely.

What actually addresses it

Naming the phenomenon. Naming impostor phenomenon explicitly — and naming its AI-specific activation — reduces its power. The first intervention is language: "what you're experiencing has a name, it is documented in research, and it is not evidence of inadequacy."

Restoring human visibility. Verification steps, authorship credits and human-in-the-loop workflow design make the professional's contribution visible. Invisible human work amplifies impostor phenomenon; visible human work counters it.

Structured competence evidence. Outcomes attributed to specific human judgements — the professional's selection, verification and contextualisation, not the AI's output — build evidence against the impostor narrative.

Psychological resilience training that engages the AI dimension directly. Generic confidence-building does not address impostor phenomenon responses specific to generative AI. Training that names the mechanism and provides frameworks for engaging it reduces anxiety more effectively than either psychological training that ignores AI or AI training that ignores psychology.