“Robots will be able to do everything better than us”—Elon Musk, 2017
As an author and editor, I miss the good ol’ days when publishing was framed around distinct roles:
Roles varied but had defined parameters, required unique skills, and were compensated based on each skill set.
Today, however, many job posts combine these functions under a single umbrella. A proofreader opening, for example, lists duties that include editing, copyediting, and fact checking. Or require SEO, HTML and XML knowhow, treating the job title like free beverage refills with payment based on the lowest-paying task.
The result? Poor training, job apathy, inferior work quality, damaged company reputations, even lawsuits. (Remember the Oxford comma kerfuffle?) Enter artificial intelligence (AI), marketed as more efficient and more time- and-money-saving alternatives to human workers. Ouch.
What is AI?
The number and scope of AI definitions, categories, and subcategories are all over the place. I combed through several bloated definitions before finding one that’s simple and straightforward. Ironically, it’s the definition John McCarthy, computer scientist (of course!) and an AI founder, provided after coining the term artificial intelligence in the mid-1950s: “the science and engineering of making intelligent machines.”
Um, ok. Machines (computers, robots) manufactured and programmed (trained) to process information (data, inputs) and solve problems like a human would. All of this fueled by algorithms, metrics, and cognitive and behavioral cloning.
Like many technologies, AI serves good and evil agendas. "Good" uses:
Questionable, unethical, and illegal uses:
How does AI affect creatives?
AI is called on to generate creative, medical, technical, marketing, and other content including book covers, illustrations, even translation services. Paralyzed by writer’s block or deadline desperation? AI to the rescue!
Trilogy, a German medical writing and consulting firm, uses AI to produce clinical and technical content. Neuroflash, a German business analytics outfit, claims its AI Book writer can produce fiction and (perhaps) nonfiction, including business and marketing applications that are search engine optimization (SEO) friendly. Neuroflash further claims its technology “. . . may even be able to produce a better book than a human author, since it is not limited by the same cognitive biases and limitations.” (Biases in their many different forms are notinherently negative, but that’s another topic.)
In addition to linguistic, spelling and dialect challenges, translators face cultural nuances of language. Imagine the effect not only on literary translation but on legal and business translation as well.
According to Meta, “We’ve built a single AI model called NLLB-200, which translates 200 different languages with results far more accurate than what previous technology could accomplish.” The program is named No Language Left Behind. It’s not clear how the open-source program affects translators who, according to the Editorial Freelancers Association, “. . . re-create a work, published or unpublished, from one language into another, checking for consistency of tone and accuracy ensuring that the original meaning is maintained. They may also review an existing translation. Unlike interpreters, who paraphrase the spoken word, translators focus on the written word.”
Although controversial, AI text-to-image programs are on the rise, generating book covers and illustrations. As far as I can tell, there’s no universal and comprehensive regulation or oversight of AI in the U.S. publishing sector, meaning no legal standards regarding liability, ethics, or copyright infringement issues.
What about editorial workers?
I’ve worked with different kinds of copy, including legal and financial proofreading, which is tedious and repetitive. A single word or phrase that’s missing, misspelled, misplaced, or defined inconsistently can render a contract legally nonbinding.
Whether due to more human errors or tighter budgets--or both--AI software is marketed specifically as a proofreading tool that identifies potential issues with drafting, editing, and reviewing legal copy (potential, as in likely or probable). Other risk-detection suites claim to streamline workflow and communication, thereby expediting project management and reducing the number of document review passes.
What are some AI pros and cons?
AI-enabled devices--calculators, computers, virtual assistants, and chatbots or chatterbots offer immediate gratification and problem resolution. This is particularly true in factory and assembly environments that depend on repetitive actions. Overreliance on at-your-service fixes, however, can hamper development of original ideas, jeopardize comprehension and, as mentioned, create legal minefields.
I’m AI schizophrenic. On the one hand, I draw on Google, plagiarism and cliché detectors, name and title generators, even plot generators for writing help. At the same time, I worry about overdependence on them and landing at the bottom of a slippery slope populated by lazy-brain people. (Melodramatic but you get the idea.)
Editors balance creativity and discipline while maintaining the author–reader connection. (Writers, remember, write to communicate.) When fed a digital diet of algorithms, can robots
In the end, will AI become a find-and-replace tool for creatives? Will it make us what someone called the architect of our own demise?