Age of Ultron or Nah? AI and the Future of Content

If Gary Halbert is to be believed (and there’s no reason he shouldn’t be), the written word is “the strongest source of power in the entire universe.” And now, the machines have this power … or so we hear. This year has been one of extraordinary advancements in artificial intelligence, pushing many boundaries, especially in the creative fields. We’ve seen AI-generated artworks, and now we’re witnessing the AI revolution in writing – in particular, with ChatGPT, officially launched only a month ago. A large language model trained by OpenAI, ChatGPT has been designed to generate human-like text based on prompts. The model is trained on a massive amount of data, allowing it to generate text that is often difficult to distinguish from text written by a human.?

Naturally, this opens up endless possibilities in the realm of commercial writing. And being an agency that deals quite heavily with the written word, at Heltr Skeltr, we had to see for ourselves: Have the machines finally risen? Has Skynet taken over? (Films from the 2000s were REALLY paranoid about this “AI is taking over” scenario!) So tasked with a mission, I took off the tinfoil hat (it was making incorrigible noises) and followed the facts.?

When I started writing the first chaotic draft of this blog, I was tempted to echo the question that has been doing the rounds on social media: Is it time for our ilk to pack up and look for other jobs? Perhaps there is an AI out there right now, drafting a blog on the future of copy and content – or publishing it already – as I languish in deadline purgatory. Am I in a race against robotic ruminations? Is sophisticated AI slowly going to phase out human writers in the field of digital content and marketing?

The short version? Not in the foreseeable future. (Take a seat, Ultron.)

The long version? The concern is not without cause. After all, humans have always had their fears about robots hijacking their jobs at the fraction of the costs and time – first the repetitive stuff (think automation), now the creative (looking at you, DALL-E). When it comes to content writing, whether we think a sophisticated AI can deliver ready-to-publish creative content depends very much on our view of content itself.?

Do you see content as primarily a way to populate your online platforms? A way to stuff in relevant SEO keywords in hopes of generating Google hits and traffic? Are blogs something to be checked off a list of marketing strategies that “everyone is doing” and is therefore something that needs to be done out of routine necessity? Do you patch out your brand’s content-related needs to the lowest bidder in a pool of writers who may not be familiar with your particular niche, or even your industry, expecting quickly churned out template content??

If the answer to any of this is yes, then your Ultron may just be here. Here’s what ChatGPT can do, in its own words:

Sounds familiar? It should, because put together, this (kind of) describes a content mill on steroids – a set of skills that form a baseline for good content writing, but barely scratch the surface of what may count as setting a benchmark. Effective copy – whether longform brand content or snappy ad copies – is essentially storytelling and requires a lot more than simply being able to write articulate, grammatically correct sentences. Good, purpose-driven writing that engages the target audience and aims to ultimately acquire customers goes beyond great use of language or clever execution. It must demonstrate expertise and give readers food for thought – drive action. The current AI capabilities can, at best, repurpose what’s already out there. (If anything, “objectivity and neutrality” that lacks any unique opinion or take is actually detrimental to the marketing cause.)?

And as ChatGPT will tell you, here’s what it absolutely cannot do:

1. Understand and express emotions: As an AI, I do not have personal feelings or emotions and am unable to experience or express them. Human content writers may be able to more effectively convey emotions and connect with readers on an emotional level.
2. Write with personal experience and perspective: Human content writers have their own unique experiences and perspectives, which they can bring to their writing to give it a personal touch and make it more relatable to readers.
3. Use creativity and originality: While I am able to generate original content based on my knowledge base, human content writers may be able to come up with more creative and original ideas and approaches to writing.
4. Understand and write for specific audiences: Human content writers may have a better understanding of their target audiences and be able to tailor their writing to better appeal to and engage those readers.

This settles at least one part of the debate. The creative capacity of current AI is vastly limited, and in a market chock full of content, AI-generated writing will fail to stand out, a quality that is not just an advantage but a necessity for brands.

But what if all you want is accurate, clean, descriptive content (with a few jokes thrown in) at a productivity rate simply impossible for any human to achieve? We’ve all heard the “we need to push out more content” spiel – more blogs, more posts, more presence. Surely AI can meet this quantity over quality demand, without running out of steam or hitting writer’s block? Even if it is impersonal and potentially irrelevant, it should be able to replace (and in the process, save a few bucks on) humans for the end-to-end production of thin, low-quality content?

Not really. (Bad news for anyone who was gearing up to make some budget cuts.)

Why? Because even a tool as impressive as ChatGPT is unable to ensure one critical baseline feature: Accuracy. ChatGPT is simply incapable of carrying out research or a Google search (even if you supply the sources). Just ask it, and it’ll tell you.

I do not have the ability to browse the web or access new information, so I am unable to mine data from links or perform any other internet-based tasks.

It also cannot fact check the content it generates.

As an AI language model, I am not able to fact-check or verify the accuracy of the information that I generate. While I have been trained on a vast amount of text data and can generate responses based on patterns and correlations that I have learned from this data, I do not have the ability to verify the accuracy or validity of the information that I produce.

It is important for individuals to fact-check and verify the accuracy of any information they use, whether it is generated by an AI language model or produced by a human. It is the responsibility of the user to ensure that the information they are using is accurate and reliable.

If you have any doubts about the accuracy of the information that I provide, I recommend verifying it with other reliable sources. It is always better to be safe than sorry, and taking the time to fact-check can help ensure that you are making informed decisions based on accurate and reliable information.

This lays to rest the remaining debate as well. Publishing content that does not strike a chord with your audience can be a pointless endeavour that yields minimal or no results. But publishing potentially inaccurate or factually incorrect content can irreparably damage brand reputation, costing much more than you may save on a writer. Of course, an argument can be made that humans are not above inaccuracies and errors. But the key difference is the ability to verify and take accountability.?

ChatGPT’s inability to use a search engine, data mine, conduct research, or even fact-check information means that the results can be highly unreliable, ranging from comically inaccurate to horrifyingly off base.?

Have a look (and some chuckles).

Here’s ChatGPT creating a fluff piece with zero actual information:

This makes sense, since the AI is not capable of conducting fresh research and can therefore only effectively provide responses to a limited set of fact-based queries. (For example, it does know who Bill Gates is).?

But things can take a darker turn.

Let’s take a look at another, almost identical piece on a different actor.

What do we learn? About the actors, absolutely nothing. About me, that I obviously watch KDramas. And about ChatGPT? That, if cornered, it’ll resort to template fluff (even when it may have some access to specific information) and just swap out the names, instead of spitting out a 404-type error saying that it is unable to create the content requested. (Ironically, it is a rather human quality to pretend to know something you don’t. Not one you’d want in your writer though.)

Well, at least it told us what it can’t do.

What about what it says it can? I did a quick test for a fairly straightforward translation.

Reader, this is what it made of I’m tired and should probably go to sleep.

Is this revenge, ChatGPT?

(Note: The above Google translation is, in fact, accurate, which we all know can be a bit of a wild card. Surprisingly, it even got the English to Bangla correct, where ChatGPT failed.)

So, are AI writing tools irrelevant to content writers? Should we just go about our days and wait for the hype to pass??

Absolutely not.?

AI writing has come a long way in recent years, and it holds the potential to revolutionize the way we create content. Think of it as a friend you can bounce ideas off (but who won’t throw something at you in frustration after your nth iteration). When you’ve hit a wall and can’t get the ball rolling, AI tools can quickly toss out interesting ideas or lines of copy. In writing this very blog, ChatGPT produced some interesting responses to my prompts and kept me entertained, if nothing else (which is a feat; just ask the ten unfinished shows on my Netflix). Credit where credit is due.

At the current level of sophistication, AI can be used to generate briefs or even create basic prompt-based copies that writers can build upon – with additions, enhancements, fact-checks, customization for the target audience, and so on. Companies may benefit from setting up processes that incorporate AI tools into the standard writing practice, for quicker turnaround times. It is important to keep in mind that ChatGPT is, at the end of the day, a “language model” – and writing is much more than just that. AI writing tools can be to writers what tools like Grammarly are to editors and proofreaders – and a process can be envisioned that starts with the former and ends with the latter, with humans doing the critical heavy lifting in the space in between.?

What AI offers in writing is akin to automation in cars: assistance and efficiency. Human control and supervision remains critical. Only we know why we’re going where we’re going, how we want to get there – do we want the scenic route or the shortest? – and how not to collide into oncoming traffic on the way. As writers, we need to stick around to do the research, make connections based on our own insights and ideas, and add that valuable human element to tell stories that make an impact.

Taking our eyes off the road is simply not an option.