Paper argues LLMs make poor human simulators because their memory is too reliable to mimic realistic forgetting

Integrating cognitive frameworks could help models simulate human memory decay