The Encoding Of Language
Historically, academics are prone to wax poetically about the nature of language. Dissertations abound about linguistics and the culture of words. Still, the basic construct of language never changes. At the end of the day, language is a code, and no one speaks the language of code better then computers.
Machines That Can Write
Even though writing has been a product of the Arts, computer science has muddied the waters by using artificial intelligence to create writing. Machine learning has established that word choice may be less about artistic license and more about the science of communication.
- Basic Editing. The use of spell check programs finds typos (form instead of from) and alerts to misspellings. In some cases, it even tries to offer a choice of words that you might mean. This is a basic example of a machine anticipating and guiding writing.
- Advanced Grammar. Texting and emails introduced abbreviated language and established a shorthand for common phrases. These expressions (lol, brb, idk) became a staple of keyboard writing. Programs analyzed these and other common data points, boiled down language to essential elements, and formed an understanding of composition. Text apps and email services began to make suggestions beyond just auto-correcting typos to predicting your next thought.
- Topic Development. The evolution of computer-generated language built on keyboard language and formal grammar. Programmers coded algorithms that could figure out patterns of speech and the structure of sentences to recognize context. Context requires a sense of continuity and organization of ideas to form meaning and anticipate the next possible string of words. Today’s text generators achieves this level of complexity.
Is It Affect Or Effect?
From Autocorrect to Smart Compose, word processing software has developed predictive capability to finish our sentences and understand our conversations. Whether in a text or an email, a web service or program, computers are now editing our communication and writing for us.
Writing has become a function that machine learning can duplicate, master, and perform. Humans still need writing skills—we still need to recognize good writing to approve the outcome—but editorial artificial intelligence is capable of handling the production.
A Literary Journal And Bot Poet
What does this say about the art of writing? Poetry and word craft have always been drawn from the heart and soul; can a computer really generate such works, then? Apparently.
Text generators are capable of crafting more than just a sentence or headline. In 2011, the Duke University literary journal published "For the Bristlecone Snag", a poem assumed from human origin but really created by a generator. Today, poem.exe is the author attributed to 1000’s of Twitter poems. Who knows, the next New York Times best seller or Pulitzer prize winner may be written by a bot…
