Philosophical foundations of AI-mediated inquiry: filling the gap in automated knowledge production

Purpose This study aims to examine these philosophical dimensions within Sindh’s academic community by employing a convergent parallel mixed-methods design. Design/methodology/approach A structured online survey of 120 faculty and graduate students across ten institutions (e.g. University of Karachi, University of Sindh, Mehran University, NED University) quantified AI adoption at key research stages where literature screening (85%), qualitative coding (60%), hypothesis generation (40%) and drafting (46%), and measured participants’ ontological, agency, validity and ethical-concern stances. Findings Inferential analyses revealed that the use of AI for hypothesis generation significantly increased acceptance of AI agency, that senior faculty exhibited lower trust in AI outputs compared to graduate students, and that broader AI usage correlated positively with trust. Concurrently, semi-structured interviews with 10 authors whose recent publications explicitly acknowledged AI assistance surfaced four themes: AI as an efficiency collaborator, epistemic uncertainty and trust negotiation, contested agency with researcher oversight, and ethical dilemmas around bias and data privacy. Integration of quantitative and qualitative findings highlighted a model of “hybrid scholarship,” wherein AI accelerates routine tasks, but human interpretation and moral responsibility remain central. Originality/value Drawing on these insights, this paper proposes a modular philosophical framework that delineates human and machine roles at each research stage and offers actionable guidance for training, institutional policy and ethical oversight in Sindh’s academic research.