The Human in the Loop

I was laid off because of AI.

There wasn’t a confrontation. Not even a real conversation. Just a brief meeting with HR about “shifting priorities” and a calendar invite for my yearly review that disappeared.

I was the Executive Editor at a digital marketing agency, and then I wasn’t.

For years, I managed a team of about twenty freelance writers. We wrote content for clients in every corner of the internet: discreet mail-order birth control, flashy financial tech startups, gourmet dog food tailored to your pet like a five-star meal plan. To say it was a variety is an understatement.

I had a great team. Great humans. We had people who could research anything and write about it in a knowledgeable, engaging, often funny way. They met every deadline. They cared.

Then ChatGPT was released.

That very first day, the owner of our company gathered us on Zoom and said, “Let’s see how many articles we can generate.” He fed it horrible prompts like “Write a 500-word article about why dog food is important” or “Write an 800-word article using the words ‘birth control’ and ‘Christmas.’” ChatGPT released in November, and we wanted timely pieces after all.

But those outputs were atrocious. They were riddled with misinformation, robotic in tone, and always structured the same way with a call to action at the end that sounded like a used car salesman.

At the time, we didn’t have the vocabulary for what we were seeing — “hallucinations.” And we knew nothing about “prompt engineering” or “AI personas.” I just knew I was watching something that felt powerful and wrong.

Meanwhile, my writers were emailing me. “Any new assignments? Just checking in.”

I had to lie. I told them things were slow.

Behind the scenes, my boss kept saying, “Cull the list. Who’s the weakest link? Whose edits take you the longest?”

They weren’t even sly about wanting to replace them, no matter how much I explained why we needed them. I defended their work, pointed out their strengths, reminding leadership that we built relationships with our clients because of the people behind the writing. But none of that mattered anymore.

It didn’t matter how sharp or reliable or fast they were — not if a bot could churn out something “close enough” in seconds. Quality wasn’t the point. We were creating content to feed an algorithm. The goal wasn’t to write something good — it was to write something fast, keyword-dense, and long enough to rank on Google. Substance was secondary. Humanity? Optional.

Editing AI content was another beast entirely. Every sentence was predictable. Unsupported claims, repetitive syntax, bland word choice. Each edit became a rewrite — not just fixing errors, but adding soul. I had to reintroduce humanity into something that was never human to begin with.

But the clients loved it.

They didn’t read the drafts, but they loved how fast we could generate articles. “More bang for the buck,” they said. They thought AI meant more content, faster. To them, volume was value.

To me, every edit was terrible. I dreaded seeing an article without an author. An AI draft took three times longer to clean up than a human-written piece. I’d stare at my screen and feel like I was patching holes in a sinking ship made of keyword-stuffed nonsense.

A couple months later, I watched a webinar where a different agency bragged about generating 10,000 articles with AI. The editor described reviewing that content as “psychological torture” — so she trained AI to edit the AI-generated content. That moment stuck with me, not just because of the absurdity of it, AI editing AI to save a human’s sanity, but for how casually it was framed as a solution. As if all of this was inevitable. As if the goal wasn’t to improve the work, but to minimize how much of it humans had to touch.

Ironically, that’s also when I first heard the phrase “human in the loop.” It was supposed to mean humans had the final say. Oversight. Guardrails. Content should have human judgement as the final filter. But the way it was used felt more like damage control, a bandage over a machine-made wound.

I didn’t feel like a human in the loop. I felt like a loop was tightening around my neck.

All this content was created to boost SEO rankings — not to inform or delight. Just to exist. Search engines reward frequent, keyword-rich updates. So companies flood the internet with junk. Articles that parrot the same advice, written by no one, reviewed by no one, optimized for everyone, but helpful to no one. We were burying the actual information in noise.

And I knew it. I knew what we were doing. But I did it anyway. What choice did I have?

I needed the paycheck. I wanted to fight for my writers, at least as much as I could. I kept thinking maybe things would shift. Maybe clients would notice the drop in quality or the algorithm would catch up. But hope like that is expensive. And exhausting.

In March 2024, Google rolled out a core update that quietly punished low-quality content. They didn’t say “AI,” but they didn’t have to. “Targeting low-quality, unhelpful, and spammy content,” — that was 90% of what we’d published in the months prior.

We pivoted. We finally gave our writers more work. But then came the questions about cost. My boss constantly complained that we were paying more again, when the machine was “free.”

Free, except for all my time. Free, except for the parts of myself I was slowly selling off.

I argued. I offered counter evidence. I explained why the AI content didn’t perform. He didn’t care. Not really.

I started therapy for the first time in my life that spring. The world was spinning too fast, and I wanted to get off. My therapist told me to try deep breathing. I never went back.

Eventually, I got better at using AI. I learned the prompting tricks — gave it a persona, told it to act like a veteran copywriter, and fed it tone and structure, pasted in examples to mimic. It got smoother. Cleaner. Passable.

But still, I’d look at a sentence and know that no one *felt* anything writing this. No one got paid for it. No rent was covered, no groceries bought, no daycare bill eased.

The writers I worked with had instincts that were worth paying for. They’d call me out if a brief was unclear. They’d ask good questions. They’d sneak in a line that made me laugh, or think, or pause.

AI doesn’t pause. It doesn’t think. It imitates what it’s seen before.

If artificial intelligence is a reflection of human intelligence, it reflects only our surface. Our hunger for speed. Our worship of scale. Our ability to optimize meaning into meaninglessness.

What makes a machine different from a human?

A machine fills a page.

A person writes something worth reading.