Artificial intelligence applications have become both beneficial to creative artists, photographers, scientists and writers, and harmful at the same time, because the protection of their intellectual property rights can be improved or infringed by these applications.
Experiments conducted by British biologist Sholto David revealed that artificial intelligence can be two or three times faster than humans in detecting copied or manipulated images in research papers.
An exploration application called “Image Twin” can scan any paper and within a few minutes it can detect any “controversial” images that could pass by the human eye, according to David’s experiments, the results of which he published in the scientific journal Nature.
David said that he devoted most of the past months to examining hundreds of research papers to discover the presence of any “duplicate images” in them.
During the experiment, the researcher used artificial intelligence to examine the papers, discovering all 63 papers that contained scientific plagiarism, and discovering the presence of more than 40 additional “questionable” papers.
But while AI applications can be useful for examining academic papers, they can do so because they have been “trained” by being fed a huge amount of content from a wide range of scientific journal websites and different content sources, according to Nick Vincent, assistant professor of computer science at Simon Fraser University. .
Vincent added in an article published by the Atomic Sciences Journal that “the producers of this content used to train artificial intelligence applications did not receive any financial compensation, nor were they even asked about their opinion on the use of these materials.”
Vincent described this development as “an unexpected evolution of AI capabilities” and leaves “content creators” without “any real ability to have a say in the use of their production to train generative AI applications.”
