A Comparative Benchmark of ChatGPT and Gemini in Literature Reviews

This dataset contains 60 peer reviewed literature reviews curated for use in ZONEDO. Each record is generated using two generative agents. GA1 corresponds to ChatGPT. GA2 corresponds to Gemini. The dataset is structured using a numeric identifier system. The final digit of each identifier encodes the academic domain of the source article. Identifiers ending in 1 or 2 represent the management domain. Identifiers ending in 4 through 8 represent the food domain. Identifiers ending in 9 through 12 represent the computer science domain. This encoding enables rapid filtering, traceability, and domain specific analysis across agents and disciplines. The dataset supports comparative evaluation of large language model outputs across domains. It is suitable for studies on content quality, reasoning patterns, and domain sensitivity in generative AI systems. Using two state of the art models strengthens internal validity and allows controlled cross model comparison, which is a standard approach in recent AI evaluation research.