Answers are grounded in this paper and cite their sources.
Background AI systems are increasingly used as emotional companions, yet their intersection with loneliness experienced within marriage remains largely unexamined. Marital loneliness is associated with relational conflict, declining satisfaction, and psychological distress, and individuals may turn to AI to compensate for unmet emotional needs within the marriage. Despite growing interest in both phenomena, no study has modelled their interplay using network-based methods, nor validated the relevant instruments in Arabic-speaking samples. This study addresses both gaps among married Saudi adults. Methods A cross-sectional correlational design was applied. 306 married individuals in Saudi Arabia completed the AI Attachment Scale (emotional closeness, social substitution, moral appreciation) and the Loneliness in Intimate Relationships Scale (detachment, emotional hurt, guilt). Confirmatory factor analysis evaluated instrument validity. Exploratory Graph Analysis (EGA) with graphical LASSO estimated network structure, and eight centrality indices identified the most influential nodes. All analyses were conducted in R version 4.5.0. Results CFA supported acceptable fit for both scales. EGA identified two communities corresponding precisely to the two constructs (density = 0.67; all edges positive). Emotional hurt was the most central node across all indices (Strength = 0.97, Eigenvector = 1.00). Cross-community edges were weak, though AI social substitution showed the strongest links to marital guilt (0.099) and emotional hurt (0.077). AI moral appreciation was notably peripheral across the network. Conclusions Emotional hurt appears to organise the marital loneliness network, suggesting that addressing it may reduce broader relational distress and AI-compensatory behaviour. Social substitution with AI seems to reinforce rather than relieve marital pain. Clinically, couple therapy may benefit from prioritising emotional repair over directly targeting AI use.
No takes yet. Share an insight, caveat, or question.
Elsayed et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: