An analysis of overused terms by AI chatbots suggests that hundreds of thousands of academic papers may have been generated with their assistance. Researchers identified 454 specific words frequently utilized by large language models, estimating that 13.5 to 40 percent of biomedical article abstracts are either fully or partially AI-generated. With around 1.5 million papers indexed annually, this could imply at least 200,000 papers containing AI input. There are instances of overt acknowledgment of AI use in scholarly writing, while some potential AI contributions may be less apparent.
Researchers identified 454 overused terms from AI language models, revealing that 13.5 to 40 percent of biomedical article abstracts were likely generated or assisted by AI.
Approximately 200,000 academic papers published yearly may involve AI assistance, based on the analysis of language models' tendencies to overuse specific words.
Collection
[
|
...
]