Zhidong January 19 news, according to foreign media reports, recently, AI language processing tools "tl; Dr Papers" was highly recommended by scholars on Twitter. “tl; Dr papers "use the cutting-edge technology in the field of AI language processing, summarize the views of the paper according to the abstract of the paper, and can refine the incomprehensible academic articles into concise and popular short sentences, enhance the interest, not only facilitate the reader to read, but also provide a new means for education and science popularization." At the same time, its summary of the article can also inspire researchers to re-examine the object of study from a new perspective.
English slang is used for the abbreviation "tl; dr" stands for "Too Long;Didn't Read", meaning "the article is too long to be interested in reading.". For academic papers that are difficult and do not want to read, "tl; Dr papers" can help you by condensing long articles into one sentence and allowing you to quickly navigate through the content.
At present, AI language processing tools are widely used in mainstream companies such as Microsoft and Google, but they are highly dependent on training data, unable to independently identify bad information, and even help spread this information. In addition, AI language processing tools often produce processing results that violate basic common sense, and their accuracy needs to be improved.
First, a sentence summarizes the difficult paper, and the new popular science means appear
“tl; dr papers" is an AI language processing website created two years ago by Yash Dani and Cindy Wu to help them learn more about software development. Over the weekend, academics tweeted about "tl; Dr Papers' summary of its academic papers soared in popularity overnight. “tl; Dr papers" are mainly summarized according to the abstract of academic papers, with academic terms cut down, concise language, although ignoring the nuances between concepts, but the overall summary is accurate, interesting, and widely affirmed by scholars.
Professor Michelle Ryan, director of the Institute of Global Women's Leadership at the Australian National University, has published an article on "glass cliff" pointing out that women are more likely than men to be selected for leadership roles during periods when organizations are facing crisis or economic downturns, which is inherently sexist. “tl; Dr Papers" summed it up as "a lot of women are put on the 'glass cliff,' which is a bad place," a summary that is simple and clear in meaning, and can be understood effortlessly by elementary school students.
Professor Ryan tweeted in support of "tl; dr papers”
Zane Griffin Talley Cooper, a PhD student at the University of Pennsylvania's Annenberg School of Communication, also summarized a paper on "data peripheries" that traced the physics history of materials for big data infrastructure. The paper was ultimately summarized as "Big data is stored on hard drives made of very small magnets that are mined from the ground." Cooper realized, on the surface, "tl; Dr papers" are for entertainment, but in reality it can also be applied to teaching and research, such as helping students read complex papers, and generating simplified abstracts for online journals that are easier for the public to read.
Second, see how machines understand and stimulate researchers' creativity
Jathan Sadowski, a senior fellow at Monash University's Emerging Technologies Research Laboratory, commented on "tl; Dr Papers "was interested, seeing it as a catalyst for researchers' creativity." He believes that "tl; The summaries provided by the "dr papers" tend to have "accidental intelligence," which may be a byproduct of machine learning's inability to fully understand language, but the perspectives offered by machines can help scholars who dig deep into their field to see their subjects from a new and bizarre perspective.
Sadowski believes that things like "tl; AI language processing tools such as dr papers can tap into their positioning as "catalysts for creativity." Brian Eno and Peter Schmidt once created a set of cards, "Oblique Strategies," each with a motto or comment designed to help artists (especially musicians) break down creative barriers by encouraging lateral thinking. “tl; Dr Papers"can also provide a similar service, inspiring scholars to develop new ways of thinking. In fact, some companies have sensed the potential of AI in this regard, launching AI creative writing assistants.
Randomly generated aphorisms on the Oblique Strategies website
Third, limited to databases and statistical methods, AI language processing tools are difficult to distinguish between right and wrong
Current AI language processing tools are based on training data, which is usually just a large amount of text collected from the Internet, so these AI language processing tools may emphasize bad information, such as repeated foul language involving racism and sexism, and may also be biased in a more subtle way.
The inaccuracies of AI language processing tools are also a cause for concern. These tools use statistical methods to process language, understanding content differently than humans do, which can lead to some very basic errors and can even be life-threatening. Google Search has provided misleading medical advice in response to the question "Seizure Response." Last December, Amazon's voice assistant Alexa guided a child to plug the phone charger into half of the wall outlet when a child asked him to randomly recommend a challenge, and then touched the exposed metal head with a coin.
Users posted screenshots of misleading information given by Alexa
All this illustrates the weakness of AI language processing tools in the structural model, that is, the data that is the basis of the model is mixed, and there is also a gap between the method adopted by the model and the way humans understand the language. Sadowski believes that AI language processing tools like these should be handled with care so that they can play as positive a role as possible.
Conclusion: AI language processing tools expand the scope of application, but the accuracy needs to be strengthened
AI language processing tool "tl; dr papers" provides accurate and concise summaries of difficult academic papers, so that academic content is presented to the public in a more interesting form. These generalizations are not only entertaining, but also promote the popularization of knowledge, stimulate the creativity of researchers, help teaching and research work, and show the broad application prospects of artificial intelligence language processing tools.
AI language processing tools have long adopted a statistically-based empiric approach, and this language understanding method, which is very different from that of humans, has helped people make great progress in computer language processing. In the future, it is worth looking forward to whether this method can further eliminate people's concerns about its error tendency and accuracy.
Source: The Verge