AI generated summary of Pubmed citation/abstract with PMID 38091290

https://pubmed.ncbi.nlm.nih.gov/38091290

Large Language Models GPT-3/text-davinci-002, GPT-3/text-davinci-003, and ChatGPT show variable accuracy, instability, and a yes-response bias in identifying grammatical and ungrammatical word patterns, differing significantly from human performance.