In my job, I’m expected to use a variety of tools to ensure accuracy, word count, compliance, style – adherence to a host of things that keep me “in line” with the company’s overall design and standards. This causes me to wonder: as part of a team of software developers, could my team design software that automates the process of writing the documentation for their own software with enough accuracy that the documentation specialist goes the way of the dinosaur?

I mean, the whole point of some of my products is to automate processes that human beings used to perform, and to automate them to such a degree of precision that people are hardly required to be cognizant, let alone present, for the actions these programs perform. We’ve designed such reliable systems that banking, health care, military information and community design can count on big data to gather and maintain the necessary materials to run our daily lives, and to store that data, to anticipate problems before the occur, and to rectify those problems with limited human intervention.

When you think, “but this type of automation cannot be applied to writing…writing requires critical thinking and analysis!” You would be correct. But you would be overlooking tools like the lexical analyzer Wordsmith, and the automated writing tool also named Wordsmith. Created by Automated Insights, Wordsmith is the API responsible for turning structured data into prose – it literally takes baseline information and makes an article. Could I be out of a job?

Feed me some data, and I write articles, too, only you have to pay me and occasionally socialize with me. Not so for Wordsmith.

The freakish thing about Wordsmith is its accuracy. I’ve studied a good bit about semantic language interpretation, and in my graduate program at Carnegie Mellon University, I dabbled in software interpretation of language, working with a pretty notable team on designing huge dictionaries of strings of language. The thing is, computers are great at reading – they can read at much faster rates than humans, they can digest huge chunks of information and store that information at infinitely larger capacities than the human brain can and their recall is spectacular. Skeptical? Just watch the amazing Jeopardy matches between IBM’s Watson and you’ll soon see that computing power can be harnessed to cull through the informational equivalent of roughly one million books per second. Humans just can’t keep up. Humans who write can’t touch that.

If computers learn a perfect formula for the Great American Novel, we are doomed.

Something to consider. Let’s try to keep this a secret from my bosses, shall we? My team of very excellent software developers may decide that this is a project worth undertaking, and the next thing you know, it won’t be just data that Wordsmith will be analyzing.

In the meantime, I will rely on the human eye and the need for context clues and interpretation that Wordsmith and Watson lack. I’ll count on the systems of reliability and emotion. I’ll count on what I know. What I learned in that high-functioning graduate analysis: A computer cannot tell the significant difference between these two exchanges:

Scene 1: A funeral home. A somber affair, all is quiet. A man says to a woman:

“Sorry for your loss.”

The appropriate response? She shakes his hand and nods, quietly.

Scene 2: A soccer game. A sunny afternoon. The breeze is blowing gently.

A boy says to a girl:

“Sorry for your loss.”

The appropriate response? She high-fives him and replies, “No sweat! Let’s grab some pizza! Woooo hooo!” As they tumble into a minivan, shouting jubilantly, kicking off their shoes.

No computer can decipher the differences in – “Sorry for your loss.”

Sorry, Wordsmith.

I’m going out for pizza.