Scientists have noticed that Potus AI, like some human writers, tends to repeat certain words often. Now, they use this habit to know when researchers secretly use Amnesty International to write academic papers.
For example, if there is a sheet of using words like “Ganned” or “prosperous” a lot, it may raise a red mark in which artificial intelligence participated, such as how the student can overuse luxury words to look more intelligent on an article.
According to the New York Times a report, Ticket From the University of Tübingen at the University of Germany, I found that AI Chatbots uses 454 specific words, such as “comprehensive” or “complex”.
By analyzing these, the researchers estimated that 13.5 to 40 percent of summaries in biomedical magazines – summaries of research papers – were fully written or partial by artificial intelligence.
With the publication of about 1.5 million sheets annually on PubMed, which is a database for medical research, this indicates at least 200,000 sheets that can include artificial intelligence, and this number may be low if the authors have edited the work of artificial intelligence to hide its use.
Some researchers do not bother to hide the use of artificial intelligence. For example, Subbarao Kambhampati at Arizona State University shared an example of X as low -quality radiology included a memorandum recognized that the paper was written by Chatbot.
Even the paper said, “I am very sorry, but I don’t have access to actual time information or patient data because I am an Amnesty International language model.”
Other cases are difficult to discover. Unless you know the phrase “renewal renewal” – the ChatGPT feature allows users to request a re -writing answer – you may not notice it slides to the respectable magazines, as the code of the blog retreated in 2013 indicated.
Another example, included a paper on MillibeDes that included fake references of artificial intelligence, such as a child invent books titles for bibliography. The paper was pulled from one database but later appeared on another with the same fake sources.
Then there is a case of a paper retreating after readers noticed that it is nonsense, including an image created from artificial intelligence of mice with comedic genital organs. It seems as if someone had presented a scientific project with a ridiculous picture and made the computer and does not think anyone will not notice it.
The issue becomes more difficult because some academics are now changing how to write to avoid doubt, and avoid favorite words of artificial intelligence such as “Defelve”, according to Camphahamati in the New York Times.
Tübingen University researchers, in the study of scientific progress, indicate that if these results are correct, the role of artificial intelligence in academic writing can have “an unprecedented influence on scientific writing on biomedical research, bypassing the influence of major global events such as printed epidemic.”
Dmitry Kobak, the co -author of the study, finds this trend suddenly. “I think something is important like writing a summary of your paper,” it will not do it. “It is like confidence in a calculator to write your wedding covenants – you expect more care for something very important.
