Extended Abstract

Introduction

Generative artificial intelligence (GenAI) is increasingly becoming part of adolescents’ social knowledge environment rather than merely a tool for retrieving information. Through conversational systems, adolescents can ask about social relationships, justice, inequality, identity, institutions, and future possibilities and receive coherent, context-sensitive responses. In this study, social reality refers to the interpretations and judgments adolescents construct about social actors, relationships, institutions, norms, inequalities, and the social future. Such understandings are formed and revised through lived experience, interaction, and comparison among multiple sources. GenAI adds a new communicative and epistemic mediator because generated responses may not only provide information but also frame and interpret social issues. Existing research has focused mainly on AI literacy, educational applications, misinformation, trust, and cognitive dependence. Less attention has been given to how adolescents engage with generated responses on social issues, assess their credibility, and judge situations in which sources agree or conflict. Accordingly, this study examines adolescents’ engagement with GenAI responses and their process of judgment regarding social issues in Karaj.

Methods

The study adopted a qualitative approach and thematic analysis. The research field was Karaj, Iran, and data were collected in 1405 SH. Participants were 23 adolescents aged 15–18 with direct experience using GenAI systems. Purposive, criterion-based sampling was used, with attention to variation in age, gender, socioeconomic background, and patterns of GenAI use. The sample included adolescents across all four age groups, with variation in gender and socioeconomic background. Data were gathered through 23 face-to-face semi-structured interviews lasting 30–45 minutes. The interview guide addressed engagement with social issues, generated responses, credibility assessment, and judgment under source agreement or conflict. Interviews were recorded with consent and fully transcribed. Analysis followed Braun and Clarke’s six-phase procedure and was conducted primarily inductively at the semantic level. No specialized qualitative analysis software was used; coding was conducted manually in Microsoft Word. The conceptual framework integrated the social construction of reality, mediated construction of reality, epistemic vigilance, and epistemic agency, functioning as a sensitizing guide rather than imposing predetermined themes. Trustworthiness was strengthened through comparison with raw data, selected member checking, peer review, documentation of analytical decisions, and attention to discrepant cases. Ethical principles included informed consent, confidentiality, anonymity, and the right to withdraw.

Findings

The analysis yielded five interconnected themes: “dialogic interaction and cognitive reframing,” “expansion and revision of the social horizon,” “grounding understanding in lived experience,” “multi-criteria and cross-source credibility assessment,” and “situated judgment and preservation of epistemic agency.” Dialogic interaction enabled adolescents to clarify ambiguity, request simpler explanations or examples, and organize fragmented knowledge. Generated responses could broaden social horizons by introducing alternative perspectives and prompting revision of previous assumptions. However, these responses did not occupy a fixed or authoritative position; their meaning depended on prior knowledge, lived experience, and other sources of social knowledge.

Lived experience functioned as an interpretive context rather than a final criterion of truth. Adolescents compared generated responses with experiences from family, school, neighborhood, and everyday life, and discrepancies often prompted further questioning. Credibility assessment included source transparency, traceability of evidence, internal coherence, confirmation through independent sources, acknowledgment of uncertainty, and balance across viewpoints. Fluent language or confident presentation alone was not considered sufficient for acceptance.

Judgment became particularly visible when generated responses converged with or conflicted with other sources. Convergence generally increased confidence without producing absolute certainty. Conflict could lead to suspension of judgment, situational weighting of sources, consultation with trusted people, cross-source comparison, integration of more credible elements, or provisional judgment. Within-group comparisons suggested that explanation and example-seeking were more prominent in some accounts of 15- and 16-year-olds, whereas multiple perspectives, structural factors, multi-source credibility assessment, and keeping judgment open appeared more prominently in some accounts of 17- and 18-year-olds. These patterns did not constitute fixed age boundaries, and no clear recurring pattern was identified by gender or socioeconomic background. The integrated interpretation was “situated epistemic agency,” referring to adolescents’ active but context-dependent role in questioning, assessing, comparing, suspending judgment, and revising their positions.

Conclusion

The findings show that adolescents’ engagement with GenAI responses about social issues is neither passive acceptance nor complete independence from technology. In terms of the conceptual framework, social judgments are grounded in socially constructed and lived contexts, while GenAI operates as a mediated source that participates in framing issues without determining them. Epistemic vigilance appears in adolescents’ attention to evidence, consistency, source comparison, and uncertainty, whereas epistemic agency is expressed through questioning, conditional acceptance, suspension of judgment, reinterpretation, and revision. Adolescents’ judgment can therefore be understood as a contextual and revisable process in which generated responses are evaluated within a broader ecology of lived experience, prior knowledge, human relationships, and media sources. The interpretive outcome, “situated epistemic agency,” highlights adolescents’ continuing responsibility for judgment amid multiple and sometimes conflicting sources. Educationally, the findings support moving beyond technical AI skills toward critical engagement through questioning, evidence requests, uncertainty recognition, and source comparison. Schools and families can strengthen this capacity through critical dialogue rather than unconditional acceptance or complete prohibition. The findings are limited to a qualitative study of 15–18-year-old adolescents in Karaj and are not intended for statistical generalization. Future research can compare other contexts and combine interviews with direct observation of adolescents’ actual interactions with GenAI systems.