Standards for the use of Large Language Models medical diagnosis

Mihailo Svetozarevic, Isidora Jankovic and Sonja Jankovic

Introduction
Artificial intelligence (AI), by definition and in the broadest terms, represents intelligence
exhibited by computer systems. The main goal of AI is to enable computers and machines to
mimic human cognitive functions. In other words, it aims to simulate human learning,
comprehension, problem-solving, and critical decision-making. AI approaches human
cognition in two distinct ways: the symbolic and the connectionist approaches [Esteva et al.,
2019]. The symbolic approach seeks to replicate human intelligence by analyzing cognition
independently of the biological structure of the central nervous system, while the connectionist
approach aims to create neural networks that imitate the brain’s structure. To fully harness the
potential of AI in healthcare, a systematic approach to its evaluation and benchmarking is
crucial to ensure that AI becomes a positive contributor to health systems.