Cayley Jones
When OpenAI’s GPT models burst onto the scene a few years ago, fueling the enduring AI frenzy, I was apprehensive to say the least. While most people I knew were eagerly jumping on to the LLM-powered bandwagon, I couldn’t help but wonder whether us writers would inevitably be replaced by the very technology we were working with. And for a while, I was convinced that if I relied on the likes of ChatGPT, I’d be compromising my credibility as a writer.
Fast forward to today - where third and fourth iterations of LLMs exist - and not only is my job as a UX writer still safe, but I’m avidly leveraging AI-powered tools in my role almost every day.
So, what changed my mind?
After accepting that Generative AI and its LLMs were here to stay, it became obvious that if I didn’t start to embrace AI – and embrace it quickly – I’d get left behind. Confronted by AI’s staggering capabilities in my role at Helm every day, I became eager to find out what I could do as a writer, with AI on my side.
Having decided to team up with AI, I sought to find the perfect LLM-powered sidekick – one that would actually make my job, and ultimately my life, easier. The offerings out there are endless, and it was overwhelming at first knowing which intelligent product to turn to. (In fact, even while researching this piece, I fell into a Gen AI rabbit hole – lured in by countless free trials promising to 'revolutionise my workflow…’).
I started experimenting with OpenAI’s renowned ChatGPT models. And though these models were great for sparking ideas (in seconds) when inspiration ran dry, their responses often left me somewhat underwhelmed.
But the AI models weren’t the only ones to blame for their lacklustre performance. They were simply responding to my poorly crafted prompts. It's as though I was giving a chef a poorly written recipe, with vague instructions, and expecting them to whip up a culinary masterpiece. No matter how skilled the chef, without clear instructions, they’d struggle to produce exactly what I wanted. And it's the same with AI. Even the most advanced language models need clear, detailed prompts to generate exceptional results.
I quickly realised that it wasn’t enough to just use these AI tools. To truly harness their potential, I had to understand how to get the most out of them. Recognising my role in the process, I delved into the art of prompt-writing. By experimenting with different prompts and styles, and crafting concise, refined inputs – I really began to see the potential of AI-generated content.
Armed with my newfound prompt-writing prowess, I was determined to find an AI tool that truly resonated with me and could generate high quality outputs, based on my vastly improved inputs.
The penny dropped a few months ago when I tried Google AI’s large language model, Gemini, for the first time. With effective prompts, I’ve been using Gemini to enhance my workflow and streamline some of my writing process – from content optimisation, generation and ideation, to high-level desktop research, and even persona development. And while Gemini's current capabilities meet my current needs, I’m aware that this may not be the case a few months, or even weeks, from now.
I’ve been pleasantly surprised by the increasingly impressive results produced by these LLM-powered tools, and their ability to inspire plenty of great ideas. Yet, there are still many instances where they veer off course – ignoring my instructions entirely and producing mediocre outputs that don’t help me at all. Interestingly, even then, the simple act of framing my thoughts into an articulate, focused prompt has often led me to insights that surpass the machine’s outputs.
Given the dynamic nature of Generative AI, a field still in its infancy, it’ll be essential for me to keep exploring its emerging tools and products, while remaining discerning about their output. But I’ve seen that, when adopted efficiently (and responsibly), AI-powered tools can be valuable assets to us writers, instead of threats.
That being said, I have never relied (nor do I plan to rely) solely on AI to do my job. As revolutionary as these tools are, I believe they should be used mindfully, and merely aid us in our writing process – not replace us as writers. I know that I could do my job just as well without AI, certain processes would just take longer. So, as I navigate my future in this AI-driven world, I’ll continue to harness the power of these machines, but never at the expense of losing what makes my writing inherently human.
