OpenAI’s natural language processing (NLP) tool Chat GPT has garnered increasing publicattention as one of the most well-known Large Language Models (LLMs). The health sciencesindustry remains one of the last industries to seek adoption of this new technologicaladvancement, and it is viewed with both skepticism and excitement amongst medicalprofessionals. Education is needed to debunk media misconceptions about ChatGPT, clarify thecapabilities and limitations of LLMs, and demonstrate how Synapsis AI makes LLMs useful formedical sciences.
For an industry-wide adoption of this new technology to occur, medical professionals and NLPapplied scientists must bridge the knowledge gap between Medicine and A.I. through mutualeducation, understanding, and awareness. The most important facet of this goal is to establish anin-depth understanding of what LLMs can do and what LLMs cannot do, specifically as itrelates to medicine. Mutual education fosters realistic expectations and satisfaction with due tofit-for-purpose applications.
LLMs can predict the insights and data to be drawn from free text but alone do not suffice forclinical research applications. They cannot infer or reason to solve a problem or question askedof the data. For completeness in human-like critical thinking, the questions asked of the datamust be algorithmically developed with a logic based on medical standard-of-care protocols aswell as current research by physicians and medical professionals.