In an age where social media feeds, video recommendations, and targeted advertisements are continuously optimized by artificial intelligence to capture and hold attention, the cognitive autonomy of teenagers has become a high-stakes battleground. A new research initiative backed by the National Science Foundation’s prestigious CAREER award seeks to understand how adolescents can develop psychological resilience against these pervasive algorithmic forces. Nora McDonald, an assistant professor of information sciences and technology at George Mason University, will lead the five-year project aimed at empowering young users to maintain self-direction in AI-saturated digital environments.
The project, titled “CAREER: Resilience in the Age of AI: Helping Adolescents Manage Generative AI and Personalization,” has secured $748,258 in funding. McDonald’s work directly confronts a pressing sociotechnical challenge: how generative AI and personalization systems—deployed across platforms ranging from TikTok to AI-powered chatbots—reshape the way teenagers allocate attention, make decisions, and construct their identities. By moving beyond traditional digital literacy campaigns, the research will co-design countermeasures directly with the adolescents it intends to support.
At the technical core of the inquiry are recommendation engines built on deep neural networks and reinforcement learning models that optimize for engagement metrics like dwell time and click-through rate. For the adolescent brain, which features a hypersensitive reward system but an immature prefrontal cortex responsible for impulse control, these algorithms exploit a cognitive vulnerability. The resulting dopamine-driven feedback loops—triggered by likes, shares, and novel content—can fragment sustained attention and entrench compulsive usage patterns. McDonald’s team aims to quantify precisely how such algorithmic curation alters decision-making processes and self-perception over time.
Central to the effort is the “Resilience Framework,” a novel sociotechnical model that describes the conditions under which young people can remain self-directed even when digital platforms are engineered to influence what they see, think, and do. The framework integrates insight from human-computer interaction, developmental psychology, and media studies to argue that resilience is an emergent property of individual metacognitive skills and a broader support ecosystem of peers, parents, educators, and thoughtfully designed technological tools. It reframes algorithmic literacy as a learnable competency rather than an innate trait, emphasizing that awareness of personalization mechanics can restore a sense of agency.
To ground this model in empirical evidence, the research team will employ a mixed-methods design combining in-depth interviews, large-scale surveys, and participatory design workshops with teenagers. The interviews and surveys will capture nuanced experiences—moments when adolescents feel empowered versus manipulated by curated content—and document the coping tactics they already employ. In the co-design sessions, participants will collaborate directly with researchers to prototype digital tools and educational materials, ensuring that any resulting interventions are culturally resonant and practically useful.
The grant’s practical ambitions are substantial. McDonald plans to produce validated measurement instruments capable of assessing an individual’s resilience to algorithmic influence, a gap in current psychometric toolkits. The project will also generate open-source curricula for middle and high schools, community programs that train parents and mentors, and concrete design guidelines for technologists building the next generation of AI interfaces. These resources are intended to shift the public conversation from fear-based screen-time restrictions toward empowerment through critical algorithmic literacy.
The timing of the research aligns with intensifying regulatory scrutiny of technology’s impact on youth mental health. The U.S. Surgeon General has issued advisories about social media risks, and lawmakers are debating age-appropriate design codes. McDonald’s work could supply the empirical backbone for these efforts, providing rigorous evidence of how personalization algorithms sway adolescent behavior and which interventions effectively bolster resilience. The project thus sits at the intersection of developmental science, AI ethics, and public policy.
Over the five-year funding period beginning in June 2026, the team will also advance theoretical understanding of human-AI interaction. By examining how teenagers attribute agency to recommenders—whether they perceive them as helpful assistants or stealthy manipulators—the research will generate new knowledge about transparency and trust in intelligent systems. This becomes especially critical as generative AI agents grow more conversational and anthropomorphic, blurring the traditional boundary between tool and social actor. McDonald’s integrative, youth-centered approach may ultimately redefine what it means to grow up in an algorithmically mediated world, teaching adolescents not just to withstand AI’s influence but to shape the technology and reclaim their cognitive liberty.
Subject of Research:
Adolescents’ cognitive and behavioral resilience to AI-driven personalization and generative AI systems, focusing on attention, decision-making, and sense of self.
Article Title:
Teaching Adolescents to Outsmart Personalization AI: A $748,258 NSF CAREER Project
News Publication Date:
2025
Web References:
George Mason University News: https://www.gmu.edu/news
National Science Foundation CAREER Program: https://www.nsf.gov/crssprgm/career/
References:
Not specified in the original material.
Image Credits:
Not available.
Keywords
Artificial intelligence, adolescents, resilience, personalization algorithms, attention, generative AI, sociotechnical model, cognitive autonomy, human-computer interaction, algorithmic literacy, NSF CAREER, digital wellbeing.
