Belief-Based and Socially Driven Pathways from AI Investment Discussion to Cryptocurrency Investment Intention: The Moderating Role of Perceived Financial Literacy

Retail investors increasingly discuss investments with conversational artificial intelligence (AI), yet how this practice relates to cryptocurrency investment intention is unclear. Informed by a dual-process interpretation, this study tests belief-based (return expectation) and socially driven (herding behaviour) pathways, and whether perceived financial literacy conditions the herding pathway. Cross-sectional survey data from 400 Indonesian investors (89.25% male; 83.50% aged 18–30 years) were analysed with PLS-SEM (5000 bootstrap subsamples). AI investment discussion is strongly associated with return expectation (β = 0.756) and herding behaviour (β = 0.751); both mediate its association with intention (indirect effects 0.253 and 0.202, not significantly different) alongside a smaller direct effect (β = 0.184). Perceived financial literacy positively moderates the herding pathway (β = 0.149; slopes 0.120 at −1 SD and 0.418 at +1 SD). Because literacy was self-assessed and the sample is predominantly young and male, the moderation may partly reflect confidence rather than objective knowledge and may not generalise to other investor groups. Results are robust across several checks, although a Gaussian copula test indicates possible endogeneity in the AI investment discussion–herding path. The model explains 55.8% of intention variance. Education and AI-design responses addressing crowd-following are proposed for further evaluation.