A Comparison of Large Language Model and Human Performance on Random Number Generation Tasks
This preliminary work addresses how LLMs mimic human cognitive behaviors, with potential applications in cognitive and behavioral science research, though it is incremental as it adapts an existing task.
The study tested whether ChatGPT-3.5 exhibits human-like cognitive biases in random number generation tasks, finding that it more effectively avoids repetitive and sequential patterns compared to humans, with notably lower repeat and adjacent number frequencies.
Random Number Generation Tasks (RNGTs) are used in psychology for examining how humans generate sequences devoid of predictable patterns. By adapting an existing human RNGT for an LLM-compatible environment, this preliminary study tests whether ChatGPT-3.5, a large language model (LLM) trained on human-generated text, exhibits human-like cognitive biases when generating random number sequences. Initial findings indicate that ChatGPT-3.5 more effectively avoids repetitive and sequential patterns compared to humans, with notably lower repeat frequencies and adjacent number frequencies. Continued research into different models, parameters, and prompting methodologies will deepen our understanding of how LLMs can more closely mimic human random generation behaviors, while also broadening their applications in cognitive and behavioral science research.