Using ChatGPT For Content Generation

by | Jun 23, 2023

I want to address the idolisation of AI and all sorts of misunderstandings around it. It has taken the world by storm this year, and I see how often the excitement and ideas of easily generating great content can be so seductive. Yet, it is just another form of conditioning when we believe that something else can express what we have to say better than us. This is a belief I have been de-conditioning from intensively, and my mission is to give you a nutshell of how ChatGPT and other language-based AI models work. My goal is to empower you to lean into your gifts. Expressing them through content becomes so much more fluent with practice.

I did not fancy myself as a writer until I caught myself as I typing a very long post on Facebook. Suddenly, I felt this yummy taste in my mouth as I thought, “Ooooh, this is GOOD!” With this thought, I realised how much better a writer I already was – compared to how much credit I was willing to give myself for it. We often are so much better when we express ourselves from the heart, energetically lit up, not trying to say “what they need to hear”.

But with the fast-growing popularity of tools that sounds so bloody convincing (at first) – there is a huge temptation to give your power away and watch it spit perfect-looking text in seconds, saving you lots of brain space and energy. And this is where the danger happens.

It is exactly the same reason why we struggle to focus when consuming digital materials. Our brains are always so overloaded with information, notifications, stuff on the sidebars, and taps open, to name just a few things. And then there are cute puppy reels… And this is so so confusing for the brain, which is a million-year-old evolving creation that is much slower in realising that there is no imminent danger of a sabre-toothed tiger jumping out of the bushes and killing us any second of the day. So the brain switches into survival mode, which means it is very careful where it allows the calorie burn. Watching, thinking, processing, creating burns so many calories, and in that overwhelming moment, the brain refuses to allow it.

This is why we feel reluctant to work hard, easily turning into anything that will “make our lives easier”, and now the 21st century is giving us “the future” – AI-based writing tools. Naturally, the excitement and adoration go through the roof.

I am not a hater. I did as well fall into that trap. A LOT! In fact, I had generated a few bits for my book, “On Marketing & Human Design: Embody Your True Self & Reclaim Your Power For Magnetic Online Visibility” at the start too. It felt so awesome. I was giving “so much value”… Only to end up deleting EVERYTHING AI generated during the edit of the first draft. You see, after a little break from being elated with the results, I saw clearly how bland, wrong, and off-point it was, even when it gave me perfectly formed sentences. And I was very frustrated with myself because I was told by influencive people that “you can train” Chat GPT to give you exactly what you need. I ended up feeling stupid for not being able to train it. Thankfully… my daughter, Chantelle, mentioned a video she saw that explained exactly what ChatGPT is designed to do, and it became so clear to me what it can be used for and where we should NEVER allow its input. If you are interested, search for a video called “ChatGPT Explained Completely” by a quirky gamer called Kyle Hill.

But to save you time, I will give you a very bottom-line explanation right here.

ChatGPT explained

ChatGPT, known as a Narrow AI, is the most common kind of AI today. You use some form of Narrow AI on your phone alone daily. As the name suggests, this AI is designed to be highly focused on one single task or set of tasks, with some optimization and tweaking allowed. It is also considered “weak” AI because it cannot do anything else outside the designed task. So what is the task that ChatGPT and similar tools are performing? In a few words: all it does is match a sequence of words based on how probable they would be used together in the desired context. Every time it accesses a massive database of WORDS uploaded based on everything digitalized before a date in 2021 (for Chat GPT), books, articles, and other internet content. Let me emphasise not information – only WORDS. So this AI does not understand or even know it is giving information because all it is doing is calculating the best “sounding” sequence of words to complete the task based on your prompt.

It does not go “search on the internet”. It does not even search for the answer in the database – simply because its task is to match the most probable used words to create the output. This is why it sounds so convincing. EVEN when the output says, “I researched on the internet” or “I searched through my database”, the AI does not even know it is saying that to you. Those are simply commonly put words to start a sentence, especially if you ask it a question. The difficulty is that often the outputs do sound like legitimate replies…

What does this mean?

All you can rely on in the output is word order. NOT the meaning, not the information, not facts. Any “correct” answers are purely coincidental. It does not know what “information” is. It only knows how to put together a sentence based on how commonly probable that specific words complete the sentence.

Let this sink in because when it did for me – I suddenly knew exactly where I could use the AI and where I do not want to go near it. For example, If I want to tell you why I do not recommend generating your content with AI if you feel you may not sound so good, I would write it myself. I want to make a point and express my perspective on what I have discovered.

YES, I am a 1/3 – “an expert on what does NOT work” – and this is such a juicy case to present. I tried it, deep dove in, got overexcited and gave it my power, only to find inconsistencies in the output that I could not ignore. And then I jumped into another rabbit hole, trying to prove my suspicions wrong if I was wrong. Only to find that video that put all the dots in their places. And I also found a great use for such tools. Without expecting it to know anything. Because, in fact, it is an awesome Narrow AI to work with language.

How To use AI in your content creation 

For example, when I do a launch, I can record a series of videos and then ask Chat GPT to give me some suggestions on an email series. I tell it what tone I want it to be. I describe in detail what I said in each video. And then, it helps me generate an email that I can tweak and customize before adding to my launch schedule. Here is a secret: I never use ChatGPT to generate my social media posts where I want to express a point or nurture any sort of mindset-shifting, educating or nurturing content. This is MY EXPERTISE, and this is what people consume before they choose to work with me.

But I absolutely use AI to generate promotional blurbs for my articles and podcast episodes, launch emails – anything that just informs about something (like a new video, my new collaborations, or even an email about this book can easily be in parts generated by AI). I can ask it to summarise a longer text I wrote – but again, I need to be careful not to distort the context. The output always sounds so convincing it is easy not to question if it actually delivered a good takeaway or summary from what you wrote in the first place.

When it comes to my podcast episodes, I found that asking ChatGPT to summarise my transcript does not generate consistently good results. Often it misses the point because the text represents how humans talk, and some things are implied or unclear to it, so it just goes back to generating the most probable words and completely loses the plot. When I tried to summarise my Visibility By Design interview with Mari Smith (A Premier Facebook Marketing Expert and an influencer well-known in the world of marketing since 2007), the output suggested she was a small business owner struggling to market her business online. I rest my case. But at the same time, it was very decent generating possible titles for my episodes and even a pretty wonderful thank you note for my guests.

So there are brilliant options where such tools are genuinely handy to save time and energy. And uses where AI is harmful to your credibility. The choice is yours, but I feel at peace now, having educated and warned you, so that you can make an empowered one.

Juliette Stapleton

Splenic Projector 1/3
YOUR VISIBILITY GUIDE

Juliette Stapleton is a Visibility Strategist for coaches and experts. She teaches how to attract clients online, creating your OWN marketing experience that feels good, easy and aligned with who you are and brings results.

Juliette has been featured in Forbes and several major business and marketing podcasts, like SocialChatter, 365Driven, Rising Tide, iHeart Radio, and Confident Live. She is an active contributor to world-leading online publications, such as Entrepreneur.com Influencive, Addicted2Success, Thrive Global, Good Men Project and many more.

She lives in Tallinn, Estonia.

Visibility is not a sprint…
slow down to accelerate