Leveraging large language models for participatory digital personas in future-oriented design: a case study of autonomous vehicle intelligent cockpits

Future-oriented design (FOD) confronts a ‘future knowledge gap’ where users of emerging technologies cannot be interviewed today, leaving designers without empirical evidence. This research introduces a systematic LLM-driven digital persona framework addressing these limitations through human-AI collaborative intelligence. Grounded in cognitive complementarity and the Double Diamond model, the framework establishes prospective participatory modelling, enabling users to validate future representations through satisfaction-based criteria. The study implements a dual-pathway methodology combining designers' empathetic interpretation with AI systematic analysis. Experimental validation with 18 participants demonstrates effectiveness, generating 273 dimensional attributes, 120 design guidelines, and 72 visual concepts. Quantitative analysis reveals divergent yet complementary cognitive profiles: Designer-driven approaches excel in individual depth, coherence, and visualisation, whereas AI-driven approaches dominate in coverage and efficiency. These profiles operationalise a practitioner decision guide for systematic task allocation in human-AI collaborative design. Expert evaluation further confirms enhanced visual appeal, functional consistency, and future orientation. Validated through autonomous vehicle intelligent cockpit design, this research provides methodology for engineering designers to address temporal knowledge gaps in emerging technology contexts, contributing to generative intelligent design theory and practice.