The cycle that began with [YOU] on AI situates the AI moment in a long arc of human development and presents it, for the most part, as an occasion for empowerment and expansion. Twenge’s framework applies a necessary corrective: the empowerment depends on the prior developmental infrastructure, and the infrastructure for a significant portion of the generation now encountering AI was compromised before the encounter began. The effort-to-achievement cycle—the four-phase engine through which self-efficacy is built—requires encounter with genuine challenge, sustained struggle, metacognitive adjustment, and the arrival at an outcome the individual can trace back to her own effort. AI disrupts every phase simultaneously, and does so for a generation that has had fewer opportunities than any previously measured to run the cycle through school, friendship, and independent activity.
Twenge’s concept of comparison set expansion identifies the specific mechanism through which AI damages adolescent self-concept in a way that social media did not. Social media expanded the comparison set horizontally, from local peers to global peers. The comparison was still between humans; the gap was demoralizing but bridgeable; the possibility of improvement sustained effort. AI expands the comparison set vertically, from human peers to machine capability. The student who compares her four-hour essay to Claude’s four-second essay is making a comparison that is, in the developmental sense, absurd—as absurd as comparing a person’s running speed to a car’s. But the psychological machinery of comparison does not distinguish between meaningful and meaningless comparisons. It processes the data and delivers the verdict: the machine is better than you. And when the gap is perceived as unbridgeable, the motivational response is not aspiration but withdrawal—domain disidentification, the protective exit from the domain that can no longer generate positive self-evaluation.
The cycle’s account of the twelve-year-old who asks “What am I for?” is, in Twenge’s framework, not a philosophical event that emerges from nowhere but the crystallization of a process that has been underway since 2012. The iGen cohort—the first generation to spend adolescence immersed in smartphones—arrived at the AI encounter already scoring lower than any previously measured generation on agency, resilience, intrinsic motivation, and the tolerance for difficulty that productive cognitive work requires. The question “What am I for?” was not created by AI. It was made urgent by it, in a psyche whose prior conditions had been prepared, unintentionally, by a decade of digital saturation.
Twenge’s framework demands that the cycle’s democratization argument be held to a higher standard of precision. The collapse of the imagination-to-artifact ratio is real. The developer in Lagos who now has a path from idea to product is real. But access does not equal agency. The tool enables; the psychological infrastructure determines whether the enabling produces genuine development or sophisticated consumption. A generation that arrives at AI with diminished agency, diminished intrinsic motivation, and diminished tolerance for the cognitive friction that mastery requires may use the tool to bypass the development the tool appears to accelerate. The output will exist. The person who produced it will not have grown.
Jean Twenge grew up in the 1980s in a family that encouraged her to notice when things changed. She completed her doctorate in social psychology at the University of Michigan in 1998 and joined the faculty of San Diego State University, where she has spent her career doing something unusual in academic psychology: studying change at the generational level rather than the individual level. Most psychological research treats generational cohort as a control variable to be held constant. Twenge treats it as the phenomenon to be explained. Her method—analyzing nationally representative surveys administered consistently across decades—is powerful precisely because it captures the cumulative effect of environmental shifts that are too slow to appear in any individual longitudinal study but too consequential to ignore across a thirty-year span.
Her early books established the approach. Generation Me (2006) documented the rising narcissism and sense of entitlement in younger Americans across standardized measures from the 1950s onward. The Narcissism Epidemic (2009), co-authored with W. Keith Campbell, traced the cultural roots of the shift. Then, in 2012, the trend lines for adolescent mental health broke, and Twenge spent the next five years assembling the evidence. iGen (2017) presented the findings: smartphone ownership was the most temporally precise predictor of the inflection, the relationship was dose-dependent, and the mechanism was displacement rather than direct toxicity. The book attracted wide attention and substantial pushback—critics argued that the correlations were too small to explain the magnitude of the crisis, that other factors such as economic insecurity and academic pressure had also changed. Twenge engaged the methodological debates without retreating from the core finding, and subsequent research, including experimental studies by Jonathan Haidt and his collaborators, has strengthened the displacement hypothesis.
Her Senate testimony in January 2026, in which she stated that her concerns about AI companions exceeded her concerns about social media, marks the extension of the framework into the AI era. Where social media damaged social and emotional development through comparison and displacement, she argued, AI companions threaten something more fundamental: the experience of productive cognitive struggle itself. A generation whose social and emotional capacities were shaped by the smartphone now faces a technology that targets the cognitive capacities the smartphone left relatively intact.
The 2012 inflection and the smartphone hypothesis. The 2012 inflection is Twenge’s central empirical finding: the sharp, statistically unmistakable break in American adolescent mental health that coincides with the year smartphone ownership crossed fifty percent. The correlation is dose-dependent—more screen time predicts worse outcomes—and it holds across demographics, socioeconomic strata, and racial and ethnic groups. The mechanism is displacement: the smartphone did not damage adolescent brains directly. It restructured adolescent time, shifting hours from the face-to-face social interaction, unstructured play, and unsupervised activity that had historically built resilience, to passive screen consumption that provided the reward of social connection without the developmental friction that real social life requires.
The effort-to-achievement cycle and AI’s comprehensive disruption. Albert Bandura’s research on self-efficacy demonstrated that the single strongest predictor of a person’s willingness to attempt a difficult task is her history of mastery experiences—direct encounters with challenges that yielded to sustained effort. The effort-to-achievement cycle has four phases: encounter (meeting a challenge that cannot be resolved immediately), struggle (sustained engagement after the first attempt fails), adjustment (metacognitive evaluation and strategy revision), and achievement (an outcome the individual can trace to her own effort). AI disrupts every phase simultaneously: it bypasses encounter by resolving the challenge before the individual engages; it eliminates struggle by providing the output without the process; it removes the need for adjustment by delivering working solutions; and it severs the connection between outcome and effort that makes the achievement deposit self-efficacy. The student who submits an AI-generated essay has the grade. She does not have the layer of agency.
Comparison set expansion: horizontal and vertical. Social comparison theory holds that human beings evaluate their own abilities by reference to others, and that the quality of the comparison set determines the quality of the information produced. Social media expanded the comparison set from local peers to global peers—a horizontal expansion that was demoralizing but motivating: the gap was between humans, and humans can aspire to close human gaps. AI expands the comparison set vertically, from humans to machines—and the machine’s capability is not the product of effort the individual could emulate. The expansion produces domain disidentification: the protective exit from the domain that can no longer generate positive self-evaluation. Carol Dweck’s research on growth mindset adds a further dimension: AI threatens the growth mindset at its root by demonstrating that struggle is unnecessary for the production of excellent output, which makes the belief that effort produces improvement feel irrelevant.
The passivity paradox and the limits of democratization. Twenge’s generational data documents a paradox that the AI democratization argument must confront: successive generations have had access to increasingly powerful creative tools and have, consistently, reported lower creative self-concept. Access does not produce agency. The time-use data shows that as tools became more capable and accessible, time spent in active creative production declined and time spent in passive consumption increased. The brain’s energy-conservation bias favors consumption over creation in any environment where both are available; and AI, which delivers professional-quality output at conversational speed, makes the path of least resistance maximally easy to find. The decline in creative self-concept was underway before AI arrived. The technology accelerates it.