Volume 16, Issue 6 And S7 (Artificial Intelligence- In Press 2026)                   J Research Health 2026, 16(6 And S7): 8-8 | Back to browse issues page

Ethics code: IR.ALZAHRA.REC.1405.039

XML Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Zarean M, Sadidpour S S. Artificial Intelligence, Social Withdrawal, and Psychosocial Well-Being: A Review of Relational Substitution and Social Scaffolding. , J Research Health. 2026; 16 (6 and S7) :8-8
URL: http://jrh.gmu.ac.ir/article-1-3248-en.html
1- Department of Social Sciences and Population Studies, Women's Research Center, Alzahra University, Tehran, Iran
2- Department of Social Sciences and Population Studies, Women's Research Center, Alzahra University, Tehran, Iran , s.sadidpour@alzahra.ac.ir
Abstract:   (22 Views)
Background: Social isolation is among the major psychosocial and public health challenges of the twenty-first century, and has been linked to depression, anxiety, cardiovascular disease, and all-cause mortality. At the same time, artificial intelligence (AI) is becoming embedded in communication, work, education, and family life, which raises urgent scientific and ethical questions about its relationship with social withdrawal and loneliness. The evidence suggests that AI’s influence on social connectedness is neither uniform nor universally beneficial; rather, it may be shaped by system design, embodiment, user characteristics, and gender. This structured narrative review therefore synthesizes the evidence on whether, and under what conditions, AI substitutes for or scaffolds human social connection, and on whether, and how, gender has been examined in these processes.
Methods: A structured literature search for publications from January 2016 to June 2026 was conducted in July 2026 in PubMed, Web of Science, Cochrane Library, Global Health Library, and ClinicalTrials.gov, supplemented by Google Scholar. English-language empirical studies, reviews, meta-analyses, and conceptual papers addressing AI and social isolation, loneliness, withdrawal, or gender-related bias were eligible. Two reviewers screened records and extracted data independently; of 34 full texts assessed (33 from the search and one report added after peer review), 28 sources met the inclusion criteria and 6 were retained only as contextual references. Design-appropriate appraisal tools were applied by source type and used to give a provisional, design-based weighting of the evidence. Because of heterogeneity, the evidence was synthesized thematically, with primary studies, evidence reviews, and conceptual papers kept separate; no meta-analysis was performed and no protocol was registered.
Results: The available evidence, which is mostly cross-sectional, short-term, or conceptual, is compatible with a dual-edged function of AI. On the one hand, AI awareness at work was associated with emotional exhaustion and occupational disengagement (not social isolation), and algorithmic amplification of gender-based hostility has been proposed to contribute to self-censorship and withdrawal from public participation, with potentially greater impact on women; this second pathway has not been tested directly. It has also been proposed that socially withdrawn individuals may perceive AI as a lower-risk social partner, which could consolidate avoidance. On the other hand, a conversational agent was associated with reduced depressive symptoms relative to an information control in one small, two-week randomised trial, whereas a second agent showed only greater improvement among more engaged users in an uncontrolled observational evaluation (loneliness was not measured in either study); embodied social robots were associated with reduced loneliness in older adults in meta-analyses; and AI companions were associated with short-term (momentary) reductions in state loneliness in one multi-study experimental and longitudinal paper, in which subsequent real-world human social contact was not measured, in mostly short-term studies. These findings are compatible with a supportive role, but companionship and perceived social connection were examined mainly in cross-sectional or conceptual work, and whether such support transfers to human relationships was not tested. Gender was associated with pathways to AI companionship use, attitudes toward AI companions, AI anxiety, and exposure to algorithmic bias; however, no included study tested whether gender moderates the effectiveness of AI interventions. AI systems were reported to reproduce historical gender biases in professional representation, conversational design, and clinical decision support; these bias studies did not measure isolation or loneliness. Integration of AI into family life, education, and healthcare adds further layers: empathetic AI may reconfigure intimate bonds and pedagogy; whether AI educational tools lacking gender-sensitive design could reinforce the withdrawal–achievement cycle documented in boys, in a study without an AI component, is an untested hypothesis.
Conclusion: Evidence on the role of AI in social isolation is preliminary and ambivalent. Whether AI deepens withdrawal or buffers loneliness may depend on embodiment, data inclusivity, person-centred design, and recognition of gendered experiences. Longitudinal and controlled studies that measure both loneliness and actual human contact, test substitution against scaffolding directly, and include culturally diverse and gender-inclusive samples are needed; gender-sensitive design and transparent bias monitoring are further priorities.
 
     
Type of Study: Review Article | Subject: ● Artificial Intelligence
Received: 2026/10/9 | Accepted: 2026/11/15 | Published: 2026/11/18

Add your comments about this article : Your username or Email:
CAPTCHA

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

© 2026 CC BY-NC 4.0 | Journal of Research and Health

Designed & Developed by : Yektaweb