When algorithms become travel planners: Benchmarking Agentic Ai in Web 3.0 Tourism

This study aims to examine how an agentic AI (aAI) system performs as an autonomous travel planner compared to generative AI (GenAI) and Web 2.0 platforms, in order to assess whether increasing autonomy enhances efficiency, sustainability, and personalisation or merely amplifies bias and opacity. The research adopts a comparative performance analysis conducted across five travel scenarios, using identical input data for all three systems to ensure methodological consistency. The results show that the AI produces the most feasible, verifiable, and context-aware itineraries, outperforming the other systems in cost optimisation, time efficiency, sustainability, and constraint handling. By providing an empirical benchmark, this study extends existing research that has largely remained theoretical, offering practical insights into AI-mediated tourism planning. The findings also highlight key policy implications: the need for closer collaboration between public and private stakeholders, and for policymakers to enhance the accessibility and machine readability of business data, especially that of small enterprises and local providers, to foster inclusion in AI-driven travel recommendations and reduce the dominance of more visible actors.