We live in an age of remarkable technological advancements. To think that the Internet has been in existence for a mere 24 years, as of the date of this article, and that primitive forms of personal computers were first introduced only 18 years prior to that is only one example of the rapid incline in computer technology. So it was probably inevitable that we would eventually come to this juncture – when the rise of Artificial Intelligence (AI) begins to mirror the rate at which computer technology has advanced over the past 42 years. One application of AI that has gained well-deserved attention recently is Machine Learning (ML), which gives a system the capability to learn more or less on its own, without being programmed to do so by a human. The applications of Machine Learning have been nothing less than varied and amazing – from predicting earthquakes to assessing whether or not a criminal is a flight risk to identifying endangered whales and much more. All of these uses are based the accumulation of scientific data. But can machine learning be applied to something as subjective as literature? According to researchers, the jury is still out on that question.
Machine Learning has, in fact, been used to examine some issues in literature that are of interest to literary scholars: the gender of authors; identifying writing styles and mannerisms of particular authors; and even categorizing books and their genres based on vocabulary and other content. One ML study was of particular interest not only to scholars, but to a host of other readers concerned with gender equity. Results of this research were presented at a recent meeting of the Association for Computational Linguistics. The study involved a Machine Learning exercise in which 3.5 million books (both fiction and non-fiction) were analyzed, all of which were published between the years 1900 and 2008. Generally speaking, the study found that men in those books were typically described by their behavior, yet women were described based on their appearance – an insightful analysis that came as no surprise to most people, but one that was for the first time based on an actual accounting of the descriptors most commonly used by authors.
While the results of Machine Learning as it relates to literature have provided some intriguing and valuable information, the majority of ML’s usefulness has come in other fields and for other purposes. Those include the automation of manual data entry, for example. Sales and marketing professionals have benefited from ML’s studies of the most effective advertising strategies, as well as forecasting sales and identifying needs of potential customers. ML and AI are also responsible for advancements in customer service by creating virtual customer service reps – digital assistants and chatbots that are capable of identifying your problem and recommending the best course of action to resolve it. And ML has proven extremely effective in optimizing data security for large corporations. There is no doubt of the usefulness of AI and its ML application for countless reasons in a wide variety of different fields.
But unlike many other areas where machine learning has been proven to be useful and successful, literature is an art form – one that creates in the reader a myriad of emotions. Great works of literature are intended to evoke those emotions, as well as stimulate conversation, deep contemplation, and debate. While ML has, without a doubt, helped to identify stereotypes and provided definitive answers to questions for which scholars were once only capable of proposing an educated guess as an answer, ML is incapable of feeling. And that, in the final analysis, is the one area in which Machine Learning is inadequate. For no matter how hard the algorithm-driven technology tries, it can’t feel. So although it may be able to provide us with valuable information regarding literature based on the text, it can’t experience the emotions that we humans feel when we read that literature – at least, not yet.