Used with care, IdeaLens could help to gauge the extent of AI use in submissions to conferences and journals, which have become 'overwhelmed' with artificially generated content, says Rishanth Rajendhran, a computer scientist at the University of Maryland in College Park and co-author of the preprint1 describing IdeaLens. The detector works by analysing the 'outline' of a work rather than the text itself. First, it distils a text ' for example, a scientific paper ' into an outline, stripping it of style. Then it looks for patterns that are characteristic of either AI- or human-generated ideas, such as those in arguments and flow. The detector was able to spot AI-derived ideas 91% of the time. It 'surprised me that they can detect AI-written arguments' from just the outlines, says Nikhil Garg, a computer scientist at Cornell Tech in New York City, who calls the detector impressive. Garg cautions that because the tool is fallible, it should not be used to judge the human authenticity...
learn more
