This free tool, known as “TLDR” (the Internet acronym for “Too long, didn’t read”), condenses a study into a concise, single-sentence summary by focusing on the paper’s salient information from the abstract, introduction, and conclusion sections. It removes things like methodological details, which are usually summarised in the abstract.
The researchers used neurolinguistic programming techniques to create this AI-powered model, training it with a dataset of over 5,411 computer-science papers with matching summaries, some written by the team itself and others by a class of undergraduate students from the University of Washington.
They further improved the performance of the TLDR model by gathering training examples in 16 other fields.
The tool has been activated for search results at Semantic Scholar, a search engine created by AI2. For the moment, it is only available in beta for more than 45 million English-language papers across the fields of computer science, biology, and medicine.
The researchers are currently improving the software so it expands to more languages and domains in the coming months.
For Daniel S Weld at Semantic Scholar, one-sentence summaries of research papers could help scientists make quick informed decisions about which are worth further reading. “Since TLDR is summarised in 20 words instead of 200, they are much faster to skim,” he said.
Short, sweet and funny
The software is not the only scientific summarising tool that cuts through academic jargon. Academics have recently shared AI-powered summaries of their research papers that “a second grader can understand”, courtesy of the website “tldrpapers”.
Professor Michelle Ryan, director of the Global Institute for Women’s Leadership at the Australian National University, posted on Twitter the AI summary of one of her articles on “glass cliff”, a form of gender discrimination in which women are appointed to leadership roles when companies are at the greatest risk of failure.
But that’s not how AI would explain it. According to the website: “Glass cliff is a place where a lot of women get put. It’s a bad place to be.”
Dr Laura Sockol, clinical psychologist and associate professor at Davidson College, also shared the summarised version of her article “Improving Quantitative Abilities and Attitudes in Clinical Psychology Courses”.
For the machine, this research paper focuses on the fact that “when students take psychological classes, many don’t like learning about statistics”. Why so? Because “they are bad at it and are afraid of it”.
Although the summaries provided by tldrpapers have had an enthusiastic reception on Twitter, the website has been labelled “under maintenance” ever since. It seems that students will now have to go through the most important parts of a research paper the old fashioned way: by reading them.