We are living in the future. That might sound like a cliché, but when you take a close look at any aspect of the modern world you’ll see that it’s an apt description. In a single generation technologies that were only speculated about in science fiction have not only become reality, but a ubiquitous component of our daily lives. Whether we’re talking about computers or cell phones (or the increasingly blurry line between the two), most of us rely on machines to conduct our business. By building robots and programming computers to take care of menial, repetitive tasks and process large chunks of information, we free ourselves up to focus on more important work. The benefits of our advanced technologies are obvious, so the only question is what sorts of jobs are better suited for people instead of machines?
A few years ago, the answer would have been simple: people work in fields that require complex thinking and flexible skill sets, whereas machines should be used for more straightforward tasks, such as analyzing and correlating data. The reason behind this division of labor was that artificial intelligence, or A.I., was still in its infancy. Machines possessed a very limited ability to adapt to different situations. Nowadays, however, A.I. is incredibly intuitive, and as machines become more and more adaptable, they can be used for all kinds of work that was previously beyond their abilities. The Associated Press (AP) has even announced that they plan on having “stories about earnings reports” generated by machines, rather than written by human reporters.
The AP is not the first news outlet to use automation technology to generate content, but they are the biggest name in journalism to own up to the questionable practice. The reason using machines to produce articles has been frowned upon is that while A.I. has come a long way in recent years, it still has a long way to go—to put it plainly, articles generated by a computer read like they were generated by a computer. English is a far more complicated language than mathematics, and machines that can calculate vast streams of numbers still struggle to produce a sentence that reads naturally. Google even has its own set of algorithm components in place to detect computer-generated articles and screen them out of an initial search result, since they have proven so unpopular with readers.
So far, the AP is sticking to Sports and Finance for its automated articles, since those fields can involve a lot of statistics and raw data. Yet by its very definition an article is written with letters instead of numbers, and without a human mind guiding it, even the mightiest computer is unable to alter its writing style to fit a certain audience. The key to quality writing is being able to adapt one’s tone, style, vocabulary, etc., depending on the needs of the project and the prospective audience, and these qualities are sorely lacking from computer algorithms. We rely on computers in part because we desire consistency, but to date all that robot-written content seems to be able to guarantee is that it will read awkwardly every time.
by Jesse B.