“The robots are coming for our jobs!”
We’ve seen this headline before. First, they came for delivery jobs and drivers, with drones delivering for Amazon and driverless vehicles from Tesla.
More recently, knowledge workers are feeling the ice-cold robotic hand looming over their shoulders. Tech startups are cropping up everywhere, promising AI copywriting. Some claim that their products can produce results to rival the output of a living, breathing, BA-in-English-holding human.
Is that true?
As a team of creatives with a few writers on the roster, we’ve got skin in the game. And as someone looking for marketing help, you do, too. Let’s take a look at how AI works and whether or not a robot revolution is likely in the world of marketing.
Unlike a human writer, a computer has never sat on a sun-drenched beach in Hawaii with its loved one. It’s also never had a blowout fight with its mother and gotten grounded the night before the prom. So how can a machine possibly craft writing that accurately evokes the emotions humans feel in these moments?
Understanding the mechanisms of AI
AI is one of those terms that gets tossed around a lot, but our collective grasp on what it actually means is murky. A survey from Pegasystems found that while 34% of consumers thought they’d interacted with AI, 84% had.
AI is much more prevalent than we think, and that’s because most of it is not hyper-futuristic. In fact, AI underpins a lot of simple tools we already use, like email spam filters.
Even the most cutting-edge AI, like the large language models that fuel some of the AI copywriting we’ve heard so much about, still relies on human creators.
At its core, AI is designed, built, and programmed by a team of humans. AI programs are also taught and trained to behave in a certain way using data provided to them by that team of humans. That’s why we see reports of bias in some AI models–because the machine can only learn from the information provided to it.
The technology is fundamentally limited by its human creators.
If AI only had a heart
Despite inherent limitations, AI tools can still get to be pretty smart. We all remember IBM Watson absolutely smoking two of the greatest human Jeopardy contestants of all time.
Watson’s lightning-fast recall of facts and figures was superhuman. It can outpace the fastest human brains. But what about its heart?
An essential element of strong writing is the ability to connect with your audience. This is as true for novelists and poets as it is for copywriters.
You connect with an audience by talking about shared experiences. That’s how you build empathy.
Even the best large language models rely on existing information to inform their writing. And sure, there’s a lot of content out there about romantic honeymoons or difficult teen/parent relationships. Why not just toss the machine a couple of paragraphs on those topics and see what it spits out?
When it comes to great copywriting, the devil is very often in the details. A generic sentence about a honeymoon on a beach or a fight with a mom won’t capture attention. The sentence that lodges in your audience’s mind is the one that evokes a sensory, emotional, or physiological reaction that hooks the reader.
Since a robot doesn’t experience feelings, how can we expect it to capture those deeply human responses as effectively as someone that’s felt those things their whole life?
Can AI keep it honest?
Another interesting quirk with today’s AI copywriting tools is their shared propensity to fib. I gave the free AI writing tool Rytr a try, plugging in some bullet points about what FATFREE does and who we work with to see what it would generate for website copy.
Some of it was spot-on, but there were some curveballs in there, too. For example, it penned this paragraph:
“At FATFREE Studio, we use creative testing methods to quickly measure your creative ideas against your goals. This allows us to refine our strategy as needed while also reducing the time needed for creative work by up to 50%.”
In addition to being phrased inelegantly, the AI tool completely made up that stat. There was no reference to the time needed to produce work in the information I fed it.
Contrary to popular belief, we need to keep it honest in marketing. We can’t have AI creating copy that touts product features that don’t exist or fabricates a company’s backstory.
In the end, it comes down to speed and efficiency. The value proposition these AI startups tout is saving time and money by hiring a robot to write for you. But is that an accurate picture of how the tools work?
I find it much more time-consuming to edit something that’s poorly written than it is to start from scratch. While these AI-generated paragraphs might be fine structurally, they need a lot of stylistic love.
Think of it this way: Would you pay a contractor to remodel your kitchen if you’d need to retile the backsplash, swap out the appliances, and replace the cabinet doors after his team left?
AI may have a role to play in supporting human writers today, but I don’t foresee a robot takeover anytime soon.