I have an interesting relationship with AI.

When I first discovered it, I was impressed—maybe that’s an understatement—I was blown away. “Look at how technology has leapfrogged tremendously,” I remember thinking. “ChatGPT has all the answers.”

And for a while, it really seemed like it did. It responded to everything I asked, never fumbling or tripping over itself. It knew what it was saying. Either that, or I wasn’t stress-testing it enough, which, in hindsight, makes me realise how much of a noob I was.

Then I was introduced to image generation tools—specifically, Midjourney.

I was amazed by the renders it produced with each prompt.

While artists were up in arms, claiming their intellectual property rights were being infringed upon, I thought, “Surely, it can’t be that serious.” I’m not sure where we stand on that now. Who won?

I took the time to refine images, challenging myself to make them look as realistic as possible—not the little chibis or random abstract wallpapers; those had no real place in my 9-to-5.

I went all in, tossing around aperture and focal length settings, demanding AI deploy ‘studio shots’ with specific depths of field and precise focal lengths—like I knew what I was talking about. I had no idea. But it was alright.

Everyone was doing it, sharing their prompts like that helpful student who tilted his exam answer sheet just enough for others to ‘game the system.’

It felt like I was getting good at it—probably because everyone around me was just getting into it.

But I always went back to the LLMs. Maybe because they felt more engaging. Ask a question, and it would respond. Ask again, and it would continue the conversation. Back and forth, endlessly. I enjoyed that dynamic. We even wrote a story together—the first parts of a screenplay, too.

Then came the concern: “AI will take our jobs.”

While the uproar around image generation didn’t trigger me, this kind of did. After all, I’m a copywriter, and Large Language Models (LLMs—yeah, I probably should have clarified that earlier) had the potential to encroach on my territory.

There was some consolation in the realisation that, at the time, they didn’t—don’t—possess the nuance humans do… yet. They lack the insights we have. So, I leaned into the idea that “AI won’t take our jobs, but people who know how to use AI will.”

Still, I worried.

What if I got too dependent on AI? What if it became the be-all and end-all?

My hubris stood guard, ensuring I never got too comfortable passing off work I had no input in as my own.

The argument was made: AI is a tool, much like Microsoft Word or Adobe Photoshop. It’s meant to help you get things done, to simplify work.

But it made things too easy.

And it bothered me how, in the pursuit of fast-tracking work, AI might make me (or anyone, really) lazy.

Then the cracks started to show.

I’d question some of its output, and it would apologise profusely, offering an alternative. I’d question that too, and it would apologise again, offering yet another answer. Rinse and repeat.

That was reassuring. It meant I couldn’t fully rely on AI. So, I supposed, I didn’t need to worry about becoming lazy.

And yet, still I worried.

ChatGPT and its peers are designed to do the most to understand what you’re trying to say. You could type a prompt riddled with typos and glaring grammatical errors, and, like an understanding partner, Gemini and company would still respond—without even once berating you for not making sense.

(My Midjourney account stopped working for some reason, but imagine I’ve inserted an AI-generated image of a marriage counsellor here, captioned: “Find you a partner with the patience and understanding of AI.”)

And that presented a new problem.

I’d preserved the integrity of my own thoughts, ensuring I put in the work because I knew AI wouldn’t. But now, I didn’t have to get my grammar and sentence structure right—AI would fix that for me.

That scares me.

I already adjust my spoken English when interacting with different vendors and service providers, recognising that not everyone here is a native English speaker. But I always found solace in knowing that when I had to write, I’d be forced to do it right.

And that’s the thing about AI.

As long as I run anything I write through Grammarly and its besties, I don’t have to worry too much about the gibberish spewed from my fingers.

I’m invincible.

But I don’t want to be.

I want to keep the vulnerability that challenges me to work harder and do better.

So while AI may not take our jobs, if we rely too much on it, we just might lose a whole lot more.