top of page

AiRobot IRL: Is AI copywriting taking over?

Picture this: you’re a copywriter, messing around on the web for news, sifting through prospects, and stumbling upon a nightmare. A bad one... And it’s coming for your job.


Countless programs offering your services for a fraction of your rates. And people are buying it. What’s happening? AI has entered the chat.

At this point, you probably know what I’m talking about; copywriting AIs may have roots in 2000s chatbots and rudimentary spellchecking, but recently they’ve been dominating conversation. So, people are starting to wonder do businesses need writers anymore? Is AI copywriting the new SEO shortcut? Has writing become too accessible? Are us copywriters about to be replaced by bigger and better? It’s starting to feel like iRobot...

How do AI Writing Programs work?

Like other Artificial Intelligence, AI Writing Programs are machine learning-based. They’re fed source material to output something unique and informed (which feels redundant, but we’ll get to that later). If you’re wondering whether there’s more to it than those two simple steps; there’s not. But it isn’t the complexity at which AI Writing Programs function that makes a difference. Rather, it’s the ability to impersonate humanness.

Cognitive Computing, Deep Learning, and other AI functions have all been inspired by neurological and ocular processes naturally experienced in human life. AIs can supposedly connect data, recognize patterns, infer, learn, speak and see just like any one of us, giving AI a foot into the world of copywriting and copy learning.

An AI’s adaptability is its strong suit. Strong examples include:

  • Neural Network Art Generator, which produces art from selected prompts

  • Dream of WOMBO, which converts existing photos into abstract digital paintings

  • MuseNet works as an AI-generated interface for musical composition

  • Riffusion.com lets you enter any genre prompt for your own instruments to play in real-time

Even with more rudimentary AIs like captioning phones that print what is overheard on calls, there’s training and improvement daily, whether as part of their own programming or with the help of a human being pulling those proverbial strings.

Why people would want to use it

Here are the perks of AI Writing. From a business standpoint, it makes sense. With its accessibility and affordable prices, plenty leap at the chance to use it over a legitimate writer.


Less to-do in writing means no cost, no talk and no time wasted. Need an email sequence, daily posts and you’re running a weekly blog? That’s a lot on the plate for one person to cover. Can’t afford a copywriter anyway? Don’t have to.

Plus, there’s always an added stress in expressing your creative vision for another person to execute. With AI, you’re left in charge to make what you want and skip the headache. If it doesn’t work the first time, try, try again. AI fixes itself with a little help. You’re bound to get something good enough for the masses. The convenience is uncanny (and free).

You’re feeling the heat now, right? Don’t worry, there’s a turnaround coming up.

Why are people worried?

If the answer isn’t obvious, you haven’t been reading. The fear of AI programs taking over creative industries as we know them is understandable. Who would want to waste their time and energy on something that can be made on the spot? Who would pay for that when they could give fewer bucks for the same amount of bang? No sane businessperson. That’s who.

In an interview with CNBC, Jim Goodnight, AI Godfather, predicted AI’s progression to include heightened image processing by the next decade. That interview was four years ago and image processing is already in app form.

So, why would we want to battle the growth of technology and its accommodating nature?

Problems AI writing programs still present

The human touch is a hard thing to duplicate. Having tested three programs myself (CopyAI, OpenAI, AIWriter), it’s obvious AI is still in its infancy.


One of the biggest issues is fact-checking. Regardless of the amount of data entered, AI programs have a hard time grabbing credible information, let alone sourcing it. Take an educational article created in seconds for the medical field; having an AI do it means risking publishing something empirically wrong and legally damaging – or fact-checking on your own to ensure accurate and up-to-date sources.

Outside some serious safety issues, having AI copywriting sites find their own works on a slew of inputted data means also dealing with the chance of plagiarism. Unless you’re wanting to shove the entire history of writing into one program (which would probably break it), AIs only have what they’ve been given to create something new, and not every kind is designed to cite sources.

That being said, comparing different programs given the same prompt also garnered similar results. Asked to write a blog about migrating sea turtles, only AIWriter gave more than a generalization seemingly written by the same person who loves bad intros. Add this to the difficulty of shifting tone per brand (another issue AI hasn’t fully countered yet) and you’re left with carbon copies for a number of different clients. A.K.A you’ve got no brand voice.

Then, there’s the issue of grammatical errors. Despite AI’s perpetual learning, it’s not without fault.

Given a prompt for a product description for a low-carb, vodka seltzer, CopyAI suggested grammatical corrections but wouldn’t automatically apply them.


Why didn’t it end a sentence with any punctuation? Why did it recognize my use of “zazz” in one generated choice but not another? Why is it asking me if it should do that? Without the human touch, one brand would be stuck with a product uniquely sold as “non-alcoholic, vitamin-infused, better than their teas and offers that sweet zawaaaaahhhhzzuuuuuurrrp.”

Options for an Instagram post weren’t much better, sometimes completely missing key elements of the original prompt: “Mountain biking in the Himalayas. One of the most impactful experiences in my life. So grateful for the people I've met on the journey, taking care of me in a land that's new to me but home for them.”

With a bland take:

“Biking through the mountains of Nepal and India. Every day was a new adventure, never knowing what was around the next corner.”

Or changing almost nothing:

“There’s no place like home. Mountain biking in the Himalayas is one of the most impactful experiences in my life. So grateful for the people I’ve met on the journey, taking care of me in a land that’s new to me but home for them.”

By the end of a one-hour session, the only thing I’m certain of is how much work remains to be done for a good final draft. Step outside the realm of general writing, and you’re faced with the issues of B2B, SEO, and other concerns. It’s tricky business getting something like an email past the spam wall – a short Instagram post only has one layer to worry about.

What AI is good at, so far

AI writing’s strong suits are giving you the first draft and sticking to short copy. Social media posts are easy print if all you need is one sentence or even one word. It’s the simplest type of writing to create and usually doesn’t require editing.


Anything more than a paragraph generated by AI, and you’re in first-draft territory... Which can still be a blessing as long as you’re willing to read over it. AI can be left to develop that 3,000-word blog (given higher word limits), while someone else checks it afterwards.


Handing off standard writing tasks is positive, let's be frank. AI in general frees up time for real copywriters to work on more complex projects. AI can take the product description, so the copywriter can plan out that commercial script (stage directions and industry rhetoric included – things AI isn’t always privy to).

As Goodnight said, humans need to manage AI, unless you're happy publishing preschool-level quality. As a former AI transcriber (I trained voice recognition software), Goodnight is more right than you might realize. AI writing works as a first draft only. There’s a strong chance that anything produced will be riddled with typos, missing punctuation, or just straight boring. The robots are still learning.

So, is AI taking over copywriting?


No, though it could, in the future.


Want to ask Amara Murphy about her time as an AI transcriber or get more details on her research? Follow her on LinkedIn.

109 views

Comments


bottom of page