My Current Feelings About AI

Honestly, the AI conversation online feels like two loud camps yelling past each other, and everyone in the middle just sighs and scrolls.

On one hand, I use LLMs all the time. Not in some “revolutionizing the world” way–just the normal stuff engineers do. Writing boring glue code, debugging something that would’ve taken me 30 minutes of digging through logs, bouncing ideas around when I’m stuck. It’s useful. Sometimes it’s really useful.

But at the same time, there’s a big part of me that can’t ignore the issues. The training data question keeps nagging: a lot of this stuff comes from scraping the internet, including artists’ work and code people didn’t explicitly agree to have used. That part isn’t just a footnote–it matters. It makes using the tech feel a bit uneasy, even if it’s helping me get work done.

Then there’s the energy question. The infrastructure behind these models is enormous–data centres consuming staggering amounts of power and water, often in regions already under resource pressure. It’s hard to feel good about a productivity boost when you know the environmental cost is being quietly swept under the rug.

And lately, there’s something else that’s hard to shake: the layoffs. Big waves of people losing jobs as AI starts doing tasks that used to fall entirely on them. Even if you’re not directly affected, it changes the atmosphere. A lot of engineers feel unsettled–like the ground is shifting under their feet. You’re using these tools every day, but also watching the impact ripple through your own field. That tension is real, and it’s hard to ignore.

Most engineers I know aren’t strictly “AI good” or “AI bad.” We’re using it, benefiting from it, and still uneasy about parts of it. “This is saving me hours” and “I’m not comfortable with how it was trained” can both be true in the same brain. But online, nuance doesn’t travel well. If you say you use AI, some people assume you support everything about it. Criticize it, and others assume you’ve rejected it entirely. It’s like there’s no room for anything in between. Justin Cox wrote a great piece on this arguing that holding two conflicting views about AI isn’t weakness–it’s the only honest position.

We don’t really have a clean language for that middle ground, and that’s part of the problem. It’s not about being “neutral” or avoiding the issue. It’s just… acknowledging that AI exists, it’s already here, it’s useful in some ways, harmful in others, and we still have work to do figuring out how to handle it responsibly.

That’s tricky because engineers are usually good at tradeoffs: latency vs. accuracy, cost vs. performance, abstraction vs. control. But when the conversation shifts to AI in society, it often stops being about tradeoffs and starts feeling like an identity test. Suddenly, a nuanced take gets read as fence-sitting, and everything turns into “pick a side.”

And that’s a shame, because most of us are somewhere in the messy middle–users and skeptics at the same time, uneasy about layoffs and copyright issues but still genuinely interested in the engineering. The internet wants a simple headline: AI good. AI bad. But “here’s how I’m using it responsibly, and here’s where it still worries me” is the only honest version. If engineers want to move forward without losing perspective–or each other–that’s the conversation we actually need to have.