Computer science has impacted almost every aspect of our world, linking individual minds in a searchable format. Every field has responded to, and many have benefited from, the innovations brought by the computer and its increasing capabilities. Journalism has been no exception. Artificial intelligence has made it possible for reporters to access hugely complex visualizations of data, instantly connecting them to stories worthy of further investigation and broadening and balancing their views. With even further implications, natural language processing and generating has enabled bots to create news articles that are nearly indistinguishable from those that human journalists would write. Journalism and computer science are now intertwined, with overarching consequences for the field of journalism and its employees.
While workers in many sectors of the economy have already seen or fear the prospect of job displacement through automation, I do not think journalists will meet the same fate anytime soon. And while computers can never fully assume the journalist’s work of reporting and conducting interviews, I welcome the contributions that AI and NLP can bring to the field, aiding journalists in making new connections and assembling stories.
Following the technological revolution of the past few decades, journalists have had to respond to myriad changes and innovations. But few ever thought that a computer using NLP would be capable of writing articles for them and taking their jobs from them. And the manner in which the computers can do this is fascinating. The NLP system first uses content selection to find the information. Next, it organizes that information, makes a structure of the text and finds the correct words to discuss the information. Lastly, the system adds in “emotion” and proper grammar to emulate a human-generated article. (All emotion written in this article came from the heart.)
While this technology has been around since the birth of the computer, NLP systems have become more accurate and efficient wordsmiths: “Things looked bleak for the Angels when they trailed by two runs in the ninth inning, but Los Angeles recovered thanks to a key single from Vladimir Guerrero to pull out a 7-6 victory over the Boston Red Sox at Fenway Park on Sunday.” It is hard to believe that a computer generated this sentence, but few people can even tell the difference between a computer-generated version and a human-written article, as a study in Sweden demonstrated. I promise I wrote this article you are now reading, but would you even know if I hadn’t?
Newspapers have started to discuss the legitimacy of allowing computers to become their chosen authors, and many have decided at most to allow computers to handle data and other tedious tasks, such as summarizing, and allowing reporters to focus on assignments befitting of their qualifications. However, others have estimated that “90 percent of news could be algorithmically generated by the mid-2020’s.”
AI systems have aided journalists in finding new stories by categorizing masses of data into data visualizations. In other cases, the computer has fully taken over the job of the reporter and NLP has enabled it to generate articles fit for print. In both, the computer has provided a tool to safeguard accuracy and identify meaningful results, while providing writers additional time for creative expression. It is a developing partnership which, while young, has already begun to bear a bounty of journalistic fruit.
However, this partnership ends at a certain point. While the computer can amass an enormous data visualization set, the AI system can only go so far, and the inferences on which story to tell from the data can be discovered by the journalists themselves. Meredith Broussard, a professor of data journalism at Temple University, describes what the computer can and can’t do: “Using the vast ‘computational’ resources of the human brain, the reporter takes only moments to look at the data revealed by the system, leverage formal and informal knowledge and make a judgment about the likelihood of a story. It would require vast amounts of computing power to get the computer to draw the same conclusions; it could take years to tease out all of the subtleties of human news judgment and implement them computationally.” Even this “expert system” has its limitations. She adds, “The human brain thus becomes an efficient and complementary part of the story-generating process, aided and augmented by the computational system.”
Only with time will we see how the balance plays out between NLP systems and journalists, not to mention whether our egos will allow proper attribution in the by-line. I believe that computer systems and journalists will forge a synergistic relationship producing the best quality of journalism with new possibilities for data visualization. In the meantime, lucky for me, I will live to see another day as a columnist.
Emily Miller ’19 can be reached at firstname.lastname@example.org. Please send responses to this opinion to email@example.com and op-eds to firstname.lastname@example.org.