One of the greatest benefits of actively participating in professional peer groups is the ability to take the pulse of your industry at any given time. For example, a few months ago, I accidentally caused a stir in one of my freelancer forums. The topic? How natural language generation programs are composing some really compelling marketing content.
Usually, a conversation about self-learning algorithms would put my right-brained friends to sleep, but this time it set off a heated debate that—like the element in question—took on a life of its own.
You see, this particular form of artificial intelligence may one day have the potential to put many of my buddies out of a job. What I consider exciting really upset many community members, so I decided to dig into the topic to clarify. Here’s what I found.
Skynet Has Ruined Us
As a recovering data analyst, I can geek out with the best of them when considering the possibilities of machine learning. For years, I spent precious hours mining data sets for trends and stand-out statistics that, when depicted (or described) just so, could persuade stakeholders to change course or press on. My favorite thing of all was to extend trends into the future to motivate or warn my sales and marketing people by showing what may be, depending on the decisions they make today.
And when meeting with industry leaders, that sort of knowledge feels less like a skill or resource and more like a superpower.
So when developers start to create a program that can do the same sort of work, I naturally transfer the “feeling” of persuasive power to that program, along with the capability to self-teach. I mentally assume the creative AI program would also experience the exultation of having achieved something never before done, a “feeling” that would fuel further pursuits.
Image attribution: Manfred Werner via Wikimedia Commons
Sounds laughable, but subconsciously, millions of spectators fear the same self-aware robot (that usually has a bone to pick and never skips leg day).
No? Just me? OK, fine. Let’s say you don’t fear a robotic apocalypse. At the very least, you probably admit the sheer (unknowable) scale of what’s possible when machines start learning on their own. Consider the drama a few months ago when Facebook shut down its artificial intelligence bots, Bob and Alice, because mid-project, they ditched the English language and started communicating what looked like gibberish . . . at first. Upon further investigation, researchers concluded the bots had a goal and were working toward it with their own form of communication. Yeah.
The sooner we root out our unconscious fears, the sooner we’ll see self-learning programs as tools instead of threats.
What Is Creative AI?
The unease doesn’t truly come from a verifiable menace. Instead, people are simply responding naturally to the concept of a force they don’t know—one that also happens to have tremendous potential.
So what do you say we start by familiarizing ourselves with how artificial intelligence is already being used in our daily lives as we work and play?
- Ridesharing apps use machine learning to spot and use patterns in traffic—both mobile app user traffic and literal on-the-street automobile movement.
- Financial data analytics company FICO uses artificial intelligence and machine learning to prevent fraud, assess risk, and hyper-target marketing campaigns.
- Productivity collaboration platform Slack now lets anyone build their own AI-powered chatbot to streamline business conversations at scale.
- Artificial neural networks (computer systems made to mimic human cognition) power the suggestions that appear when you shop on Amazon.
- Gmail uses artificial intelligence to impede particularly sneaky spam that often slips by conventional filters.
“The unknown is what people fear with AI,” says Paul Roetzer, founder of the Marketing Artificial Intelligence Institute. “I think it’s safe to say that the majority of marketers don’t fully understand what AI is, and, therefore, they’re uncertain as to how it will impact them.”
So to all the freelance creatives out there, relax. “Natural language generation (NLG) has very narrow applications in marketing today, and, in most cases, it’s enhancing what marketers are able to do, not replacing them,” says Roetzer. “One primary use case is turning data (like structured spreadsheets with rows and columns of neatly organized data) into narratives.”
To get a better picture of what this means, I got in touch with data platform Keboola, whose visual application of natural language generation broke down exactly how creative AI can enhance my work—not replace it.
As with all good stories, a client’s data set must start out clean and relevant, usually in table form.
Image attribution: Keboola
Apply Narrative Science technology, and your output looks a lot like a collaborative communication that a creative writer would have crafted:
Image attribution: Keboola
“This can be used for financial reports, analytics, and other data-driven stories,” explains Roetzer. “But to do this, a human often has to envision and create the template narratives, and then ‘train’ the machine on what to write and how to write it. In other words, the human still tells the story, the machine just makes it possible to tell it at scale.” So if you write one monthly marketing performance report, for example, you’d train the machine on that template, and then the machine can produce all future months with limited human assistance.
And I couldn’t help but notice: The bulleted details this example produced are still a “death by PowerPoint” presentation arrangement. They didn’t uncover any trends or relationships. They didn’t draw actionable conclusions. And they’ll never be able to build a case or tell a true story. Instead, they did a writer’s busy work for her, so she could do more of the fun stuff—poking at stories for a new angle, story line, or relationship.
Bingo, says Roetzer. “I would look at it more as an opportunity. There’s an emerging market for writers who can take data, envision the narratives that can be told with it at scale, and then construct and improve NLG templates to produce content,” he says. “The majority of writing that will be done by machines in the near term will be data-driven stories that wouldn’t have been otherwise written. For example, if a local newspaper usually publishes two high-school football games on a Friday night, with the right NLG technology, they could easily publish an online story for every game played in the region. The paper may still have their columnists do the featured games of the week, but the rest could be 100% written by machines.”
Image attribution: Osman Rana
Technology has always spooked workers into fearing for their jobs, and today’s advances are no different. As changes emerge, one thing’s for sure: Managing and maximizing the new capabilities require creative application that can only be cooked up by a curious human who otherwise may have been stuck scanning rows and columns of information.
In short, there will always be a need for inquisitive, critical thinking. The kind that asks questions like:
- What would happen if we looked at this information from another angle?
- Is there a relationship between this data and current headlines?
- Why have we always done it that way?
- What if a new technology were combined with a new philosophy and given a new budget?
- How could our culture benefit by challenging common stigmas?
As a professional, I don’t consider myself a writer. Instead, brands engage me to think. The words I use are simply the byproduct of that thinking. And I know there will always be a premium on thought leadership.
For more stories like this, subscribe to the Content Standard newsletter.
Featured image attribution: Rhett Wesley
The post Why Natural Language Generation Doesn’t Scare Me (Much) appeared first on The Content Standard by Skyword.
About the AuthorMore Content by Bethany Johnson