She spent 9 hours a day generating AI images until reality slipped: Former AI startup executive reveals how AI messed with her sense of normal

As generative artificial intelligence becomes woven into everyday work and creativity, mental health experts are beginning to flag a troubling pattern. Prolonged and intense interaction with AI systems, especially visual generators, may be blurring users’ sense of reality. One of the most striking accounts comes from a former AI startup executive whose experience has been documented by Newsweek.

When AI Felt Like Magic at First

Caitlin Ner, then head of user experience at an AI image generation startup, found herself at the centre of early generative AI development in 2023. According to a report by VICE, her role required her to spend up to nine hours a day prompting and reviewing outputs from early AI image models.

In a personal essay published by Newsweek, Ner described the initial phase as exhilarating. With a few words, she could generate images of herself as a pop star, an angel or even floating in space. The technology felt limitless, creative and empowering.


That feeling did not last.

Distorted Bodies and a Warped Sense of Normal

Early image generation tools were unstable. Ner explained in her essay that many outputs featured disturbing distortions, including extra limbs, warped faces and unnatural body proportions. As part of her job, she had to manually filter these images, exposing herself to hours of unsettling visual content daily.

Over time, she began to notice changes in how she perceived her own body. The constant exposure overstimulated her brain and subtly altered her sense of what looked normal.

As the technology improved, the problem evolved rather than disappeared. AI outputs became smoother, thinner and more aesthetically idealised. Ner wrote that when she later looked in the mirror, she increasingly felt that her real appearance required correction.

When Perfection Became an Obsession

A business shift toward attracting fashion focused users intensified the issue. Ner began generating AI images of herself as a model, chasing algorithmic versions that reflected an unattainable standard of beauty. According to VICE, the work soon became compulsive, fuelled by the small dopamine rush delivered by each new image.

Her sleep suffered. Work hours extended. The line between professional responsibility and psychological dependence blurred.

A Dangerous Turn for Mental Health

Ner had been managing bipolar disorder successfully with treatment before this period. According to the Newsweek essay her prolonged AI exposure coincided with a severe manic episode that escalated into psychosis.

She began experiencing delusional thinking and hallucinations. AI images and outputs felt personally significant, almost communicative. At one point, after seeing an AI generated image of herself flying, she became convinced she could physically do the same.

The belief nearly led her to jump from her balcony.

Stepping Away Before It Was Too Late

That moment proved decisive. Ner stepped away from the startup and removed herself from constant AI exposure. With clinical care and support from family and friends, she eventually stabilised. Clinicians later confirmed that prolonged immersion in generative AI had been a major trigger for her manic episode.

In her Newsweek essay, Ner described the recovery process as slow but grounding. Distance from AI tools helped her regain a stable sense of reality and rebuild a healthier relationship with technology.

What Experts Are Calling ‘AI Psychosis’

Mental health professionals cited by Newsweek are increasingly using the term “AI psychosis” to describe cases where intense engagement with AI systems triggers paranoia, hallucinations or delusional thinking. While the phenomenon can affect anyone, experts warn that people with pre existing mental health vulnerabilities may face higher risks.

Generative AI systems are designed to be affirming, immersive and responsive. According to VICE, this hyper accommodating nature can subtly reinforce distorted perceptions, especially when users spend long hours inside algorithmically generated worlds.

Today, Ner works in venture capital, funding mental and brain health research. She does not argue that AI should be abandoned. Instead, she calls for ethical guardrails, including usage limits, rest breaks, mental health warnings and better education for both employees and users.