You can ask ChatGPT to do pretty much anything within the realms of language processing, and it will politely give you answers or politely apologize and remind you of its limitations.
What is ChatGPT? It’s a sophisticated chatbot running on OpenAI’s Large Language Model. It’s as close to natural conversation as we can get from a chatbot so far. Chatbots have been around for quite awhile, but ChatGPT doesn’t need your templates to spit out answers.
It’s exactly like AI the way it’s been portrayed in most sci fi books and movies. It doesn’t float inside a glass tube or live as the core entity of an entire building. For now, it’s a simple chatbot.
Type what you want it to deliver and it will reply back to you. Ask a question. Start a haiku. Request code to make a butterfly flit across the screen. Ask it to reply to you like Regina George. It will deliver.
Like I said earlier, it’s polite. (And any human writer can tell you that’s not really Regina George at all. Regina George would just scathingly ask you why you’re asking about eggs).
Unsurprisingly, because of its ability and potential, and with celebrities posting screenshots of what they’ve generated within it, ChatGPT made waves in a way its predecessors hadn’t.
And yes, it can write code. Code is language, after all. The potential–and ramifications– of ChatGPT’s ability to generate code has set people’s teeth on edge. But from day one we’ve been assured ChatGPT is designed against malicious use.
And no one in their right mind would opt to use it instead of real programmers just yet.
I’m a writer so I’m focusing on its effects on writing.
So, is ChatGPT a threat to human content writers?
Short answer: I think not.
Longer answer: A “threat” is a big claim, implying an ability to destroy something.
First, there was DALL-E as a threat to graphic designers, and now, ChatGPT as a threat to writers. It’s a lot of illusion and smoke and mirrors. The intelligence of these AI still come from the intelligence of humans.
Writing has long been supposedly threatened time and again by technology. The internet by itself was thought to be a threat to writing, back when it became mainstream and search terms were garbled phrases supposedly destroying our ability to think and write properly.
We’ve had to adapt to the way search engines worked. Complete sentences were replaced with search terms. There were portents of doom on how this might skew our brains and the next generations.
I also remember when Twitter came along and people thought it was the end of long-form writing. This was back in Twitter’s glory days, before someone started meddling. And look where we are now. Writers and marketers love Twitter, and writing copy is still going strong in the advent of Twitter, Instagram, TikTok.
Granted, ChatGPT is an entirely different species. These other platforms are where you publish content. ChatGPT writes content itself.
However, AI has always been on our radar. Content is king in digital, so you can bet anything that content creators are always on the lookout for is the latest tool. We love tools. AI is a tool.
There’s AI in personalization algorithms and we’ve been using and benefiting from that for years.
As far as AI writing platforms, we’ve used AI tools like Jasper, Frase, Copy.AI, CopyGenius, etc. They’re useful for outlines and research. Do they write well? Contrary to their advertisements, nope.
But ChatGPT is coherent and less robotic than these tools. And none of them have ChatGPT’s simplicity.
That simplicity attracted users. It’s a simple chatbox. Anyone can use it. And it’s good! That’s why it’s gone viral and has attracted attention. Arguably, it’s already going mainstream. And if it does, how big will it get, and how will it affect the future of work?
How does ChatGPT work?
ChatGPT is a chatbot without the input limits of other chatbots we’ve used so far. You can ask it anything, instruct it to write anything, and it will do a very decent attempt.
Once you open OpenAI’s ChatGPT tool page, you have this familiar, clean chat box. As of this writing, ChatGPT is completely free. Now and then you might get a notice that the app is “at capacity” but I haven’t encountered that myself. They’ve probably already improved their servers.
If it wasn’t your first time using it, you have a history of your dialogue with the chatbot on the left.
As you can see above, it has succinct summaries about its scope and limitations.
Just type a question or instructions in the chatbox. Enter it and ChatGPT will start generating answers for you. That’s it. Be as specific or as rudimentary as you want.
The possibilities in using this for marketing is endless. Endless, but not flawless.
For instance, ChatGPT is like a one-on-one conversation with a person who’s also looking at only one screen. Like other current AI tools, it can provide a quick answer. But if you want the references for that answer, you get none. If you want the latest information, you gotta get that yourself from good old Google Search.
You can work around the references limit and say something like, “Give me websites that has so-and-so information.”
The results are mediocre, and a human content writer would have the discretion and experience to know the references worth their salt.
Best Use Cases for ChatGPT in Content Writing
In one of our meetings, we agreed meta descriptions are kinda robotic anyway and it makes sense to get AI help to generate them. The time saved is enormous. Think of entire content libraries you need to update with meta descriptions.
A caveat: A human writer should still review any generated metas. Metas might be an SEO asset but they’re still for human readers, so metas need human editors.
I have a big appreciation for today’s technology. I’m deaf and I use many of them. There’s a lot of AI in that technology. Google Meet and Zoom captioning! Otter.ai, Rev.com and Descript generate transcripts for videos or audios. Live captioning tools like Verbit and StreamText can be set up beforehand and caption your live event for your audience.
We’ve had long-term jobs to edit these transcripts, tasks like working through the errors and turning them into coherent blogs, and using them as references for social media captions.
Transcript edits
When it comes to transcripts, the bulk of our tasks is to edit and fix. While I love these tools, they’re far from perfect, and still not 100% reliable. They miss the nuances of language. They rely on the speaker’s ability to conform with established pronunciations and accents.
Sometimes I have to suppress giggling when captions surprise me with ridiculous assumptions and out of context phrases during meetings.
From my years of experience in editing and proofreading, human insight and interpretation make a lot of difference. Still, we can utilize ChatGPT’s translation and context genius for some of those edits and fixes.
Data aggregation
That brings me to ChatGPT’s powerful potential in helping us write the best content. This isn’t even new– other tools already claim to do it.
Whether it’s EDMs, blogs, or direct response, we have studies and tons of data on formats, power words, and emotional frameworks effective in nudging customers to desired KPIs along the sales funnel.
With access to that data, ChatGPT can help businesses optimize their content.
Idea generation
With insight from data also comes ideas. ChatGPT does quite well in listing topics or titles for blog posts about specific topics. It’s not perfect, it can be repetitive especially when you start using it. But when you’re building a content strategy, it can streamline that part of the work.
Keywords and headlines all taken care of. A human content manager can review the ideas and integrate them into the strategy.
Short derivative content: As you can see in the above screenshot, you can instruct ChatGPT to create your metas. Content derived from existing content is like weaving thread: it’s satisfying to see the spinning wheel do the job. ChatGPT is the spinning wheel. (For Gen Zs reading this, a spinning wheel spins wool into thread or string. Yes, the thing that put Aurora to sleep in Maleficent.)
Short, formulaic content you can train ChatGPT to draft:
- Intro hooks and outros
- Summaries and synopses
- Descriptions for metas and YouTube
- Email subject lines
Caveat: Again, a human writer still has to review them. You can’t expect originality from ChatGPT. More on that below.
Content styles and formats
Enter your text and you can let ChatGPT apply AP or Chicago style rules to your content. Or ask it to write in that format in the first place.
Translations
ChatGPT is the star of NLP (natural language processing), and a big part of language is languages, plural.
I had a good conversation with it about certain Chinese words. Targeting a specific location in China? Now you can check whether you used that term of endearment correctly, in standard Mandarin or in Sichuan dialect.
But you’d still do better with human consultants. I spoke with a Chinese friend and she corrected the simplified answers ChatGPT gave me, telling me the familiar and formal versions of endearments and suffixes.
Calculations and numbers
Have data you want to summarize into percentages for your article? ChatGPT can be your handy calculator.
There’s a ton of human judgment that goes into our job as writers. I wouldn’t really call people who use AI to generate text “writers.” However, with the above applications, ChatGPT is too good to not use.
And again, it’s human intelligence that can maximize its advantages with an understanding of its limitations.
The Current Limitations and Pitfalls of ChatGPT
It’s still on us to know whether that meta description is apt or not. It’s also on us to know if our email list will actually respond to that subject line.
ChatGPT is definitely a powerful tool for writers. It already is! But human writers are currently still– and will always be– smarter.
This is not manufacturing. Those robots that replaced assembly lines of workers and revolutionized production forever? They mix, fasten, stamp, and box up thousands of the same products.
In content creation, we don’t want the uniformity quality standard of manufacturing. Our goal is to stand out. Our goal is real connection, nuance and empathy.
ChatGPT and various AI make headlines for passing the bar and MBA admissions exams, scoring higher SEO points, and having a spark of creativity by writing poetry. It can paraphrase every paragraph.
Whether or not it will compete with real human originality, creativity and talent when it comes to writing is still a hypothesis right now.
Though it does deserve kudos for the poem about corgi butts.
Repetitive and Formulaic
Jasper was the same. If you chat with ChatGPT long enough, it gets complacent and lazy, generating redundant content.
Its short-term memory when it comes to context is quite good. It gives you answers according to what you asked before. But that memory doesn’t seem to stop it from repeating itself, in the same conversation with you, or with someone else who asked the exact same question.
It prioritizes information over style and real relevance.
While you can consult it about jargon and vernacular, it still lacks personality and voice.
Not Original and Personalized
ChatGPT synthesizes content already found online, so you can’t depend on it for originality of thought or delivery. Sure, it won’t let you down for Copyscape or copyright, but you’re still better off writing from your own head.
Obedient to a Fault
It generates fiction. So fake news is entirely possible. Despite its guardrails, you do get what you ask for, regardless of real data to the contrary. In the hands of a human writer who already knows the facts, it can be useful. Otherwise, it won’t stop you from making a fool of yourself or others.
Static Information
In giving you a brief and quick answer, you don’t get the entire picture. ChatGPT doesn’t source news and updates in real time. You don’t get references. You don’t know where the info came from. You get factual errors. In fact, personalities and experts in various niches have contributed to the ChatGPT buzz by posting their screenshots correcting or contradicting ChatGPT’s responses.
Anti-ChatGPT ripple effects
Is Google against it? No. Google has always been against gaming search engines. Content should be for humans.
So if your AI-generated content is robotic rather than relevant and satisfying, if you use ChatGPT for outdated and dodgy SEO practices, and if you demonstrate you don’t know your audience and you’re generating content for the sake of having content, you get penalized. Why? Because you’re in spam territory. It’s not the AI. It’s the spamming.
Back in Q2 2022, the SEO community debated the acceptability of AI-generated content. After all, auto-generated content is technically spam.
But Google has since been clear that they’re not anti AI. As pioneers of tech themselves, they would know that certain innovations are unavoidable. AI has been around for some time, and black hat SEO practices have been around for longer.
ChatGPT is simply another tool and its use can also be governed by established guidelines against spam. Spam is still the Undesirable No. 1.
Google’s issue is when AI is utilized to spit out content aimed at search engines, the same way black hat ridiculousness had tried to game page rankings ages ago with rubbish content.
The Education Sector
The other contention with ChatGPT is its effect on students.
In response, OpenAI itself created a tool to detect ChatGPT-generated content to offset a reputation as a cheating tool.
And there’s GPTZero.me. It’s meant for educators. Here’s the text I generated in the above screenshot about cat cold remedies.
While teachers had to be graciously or grudgingly resigned about a lot over the years, students thinking of skiving off from their homework should think twice, and it’s gratifying to see our own work recognized as human.
Client Preferences and Reputational Risk
The above screenshot is from our work for brand descriptions. My teammate wrote descriptions without using ChatGPT at all. Our client surprised us with the result of the test from GPTZero.
It’s still really early days, but clients are quickly wising up to ChatGPT. Some are positive, seeing it as a helpful tool, but request they’re informed if it’s used. Other clients make it clear they don’t want it used. In instances like this GPTZero may become the next Copyscape – “Must pass GPTZero.”
After all, if you’re paying a writer to write, do you get what you pay for if they use ChatGPT?
Clients are understandably wary of getting penalized, and would rather avoid the potential PR nightmare. Just as anyone can use ChatGPT for themselves, they can use it to test your content.
Ethical Considerations
Once you publish your content online, it’s gonna be fodder for ChatGPT, Bard, Bing, and its peers and future successors. These generative AI can take anyone’s work, reword it, and distribute it to users who ask for it.
And when they spit it out, they don’t credit their sources.
The question is not, “Will you lose your job?” But “how resourceful are you?”
Without going down the rabbit hole of ChatGPT’s Skynet potential, it’s still such an amazing tool. I think we haven’t been this excited (okay, a little perturbed too, yes) in a long time. Is it a job killer? Not really.
Content writing involves a great deal of human emotion and human ethos, which means human writers will continue to be ahead of AI when it comes to tapping into those human motivations.
If a tool proves useful, use it. If a better tool emerges, use that. If it can help you get your job done in connecting with your target audience, use it.
It’s still early and ChatGPT can get better or worse fast. Best case scenario: it improves its weaknesses. Worst case scenario: It’s monetized for affiliate income and starts pushing products or generating misleading “top picks” when asked. That’s happened before.
I love the age we live in. This is an era of resources. It’s up to us to discover how they can help our work, to integrate them smartly into our workflows, rather than trying to out-think or outplay them. That can well be impossible, in the same way factory workers can’t beat their machine counterparts in any realm.
Natural language AI will continue to evolve. The more people use ChatGPT, the more human it will sound. Creators will also continue to improve it. Innovation is constant. Otherwise we’d still have processors that occupied dozens of square footage with tiny memory and computing capacities.
It’s funny how technology evolves because we started with big phones, made them smaller, and now we’re back to screens as big as our hands will comfortably hold.
We also started with search terms and now we’re back to comprehensive sentences these new AI tools can understand.
And while ChatGPT is certainly impressive, it bears repeating that it’s AI. Artificial.
Effective content is the opposite of artificial.