Trust, Support, and Adoption Intentions Towards Generative AI are Context Dependent

With the prevalence of generational artificial intelligence (AI) in everyday life, there has been a recent increase in research on individuals' perceptions and attitudes towards generational AI. However, with the breadth of use cases and applications made available with the advancements in large language models (LLMs), there are increasing signs that attitudes towards LLM-based generative AI may be context or use-case dependent. We examine the contextualised trust of individuals in AI, the support for using AI, and AI adoption intentions in a survey experiment (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{N}=306$</tex>) across three contexts (reasons): technical, mental health, and creative assistance. Participants read ostensible news articles about AI use in one of these three contexts, and completed a survey on their resultant and previous attitudes. The results show that trust and adoption were significantly higher for technical assistance, and lower for creative assistance and mental health assistance. However, support for using AI for mental health assistance was significantly higher than creative assistance, and comparable to technical assistance. Contextualised trust and support also significantly predicted AI adoption intentions, above and beyond trust scores towards AI in general. Our study suggests that contextualised attitudes towards specific AI products appear to be sufficiently differentiated from (general) AI attitudes, and instances where AI is used for technical reasons, as opposed to creative or mental health reasons, garners more favourable attitudes.