AI Burnout From Both Ends

I'm tired. I think a lot of us are, and I think AI has a lot to do with it.

That might sound strange coming from someone who uses AI every day and writes about it here. I like these tools. They've changed how I work. But lately I've noticed the pressure coming at me from two directions at once, and one of them is me.

The Pressure on the Industry

There's an idea floating around the whole software industry right now that AI makes everything instant. You type a prompt, the feature appears, and you're done. No thinking required.

The numbers back up how wide that gap is. An Upwork study found that 96% of executives expect AI to boost productivity, while 77% of employees using it say it has actually added to their workload. Almost half of them had no idea how to hit the numbers they were expected to hit.

It's an easy idea to believe, because the code part really is fast now. But AI can't tell any of us what the code should do, or how it fits into a system that has years of decisions baked into it. Somebody still has to figure that part out.

When everyone assumes the whole job got faster, the thinking time is the first thing to get squeezed. And the thinking was always the hard part.

Everyone Has a Prototype Now

The other side of it is that everyone has AI now, not just developers. Anyone with an idea can have a conversation with it and come out with a polished looking design in about five minutes.

Honestly, that's pretty great. Some of those ideas are really good! But it also means ideas can change as fast as they can be generated, and software can't always keep up with that.

Rebuilding is cheap now. A rewrite that used to take a week might take an afternoon. But cheap is not the same as scalable. Enterprise software has data, users, integrations, and other teams depending on it. Every change has to be carried through all of that carefully, and that part doesn't care how fast the code was written.

We still need to know what to build. That's the job. Building the wrong thing quickly is still building the wrong thing.

Then I Go Home and Do It Again

Here's the part I didn't see coming.

I have never written so much code in my life. Seriously. Since AI tools got good, I've been learning new languages, publishing apps, and building multi-dimensional game worlds that I can play in with my kids. If you've read my posts on FrameBeat or the Google Health sync, you've seen some of it.

And it's fun. It really is. Building a world with my kids and then exploring it together is one of the coolest things I've ever gotten to do.

Every idea I have now feels doable in a weekend. So I do it. Then I have another idea.

It Adds Up

The problem is that none of this happens in a vacuum. There's an ever increasing stack of bills. There are family obligations. There's the regular stuff of being a parent and an adult that doesn't go away because I found a fun new side project.

Each project feels light while I'm doing it. Stacked on top of everything else, it's way more than I should be doing.

AI took away a lot of the friction that used to slow me down. I didn't realize how much that friction was protecting me. When something took a month, I had to pick carefully. Now I can say yes to everything, and I mostly have.

Turns out this is a thing. Researchers at Berkeley Haas followed about 200 people at a tech company for eight months. Nobody asked them to do more. They just did, because AI made starting something so easy. They took on extra work and filled their breaks and evenings with prompts. Reading that felt a little too familiar.

I Already Wrote the Advice

Here's the embarrassing part. I've written about this before, and I sounded pretty sure of myself.

In Seven High-Impact AI Prompts That Cut My Dev Toil, I timed everything and showed I was saving 25 to 35% of my glue work. I measured the time I got back. I never measured where it went. Looking back, I think most of it went right into building more stuff.

Then in How I Use AI Time Savings to Regain Mental Clarity, I wrote that "time saved disappears unless I choose where it goes." I still believe that. I just didn't expect that I would be the one choosing to spend it all on more code.

That same post had a list of rules. One of them was "No AI after my last work block. That time is for decompression." Meanwhile, I'm writing more code at home than I ever have, and almost all of it is with AI. So that one didn't last.

Under "What I Avoid," I listed "Letting faster output become higher expectations." I was thinking about expectations from other people. I didn't think about my own. Every time I built something faster, my idea of a normal amount to build went up with it.

And I ended that post by saying "start with one small rule and protect it for a week." A week is easy. The hard part is month three, when the rule is still there and the fun new project is right in front of you.

I think the mistake was assuming boundaries were only for work. All of my rules were about the job. None of them covered the stuff I build for fun, because it didn't feel like work. It turns out my brain doesn't really care whether the code is for a paycheck or for a game world. It's tired either way.

What I'm Trying

I don't have this figured out. I'm writing it down partly so I'll actually listen to myself. Again.

When I'm building something, I'm trying to remember that fast code doesn't mean fast decisions. I try to say out loud where the thinking time goes, so it doesn't quietly disappear. When a new idea shows up, I try to ask what problem it solves before I start building.

At home, I'm trying to finish things before starting new ones. And I'm trying to remember that the time with my kids is the point. The game world is just the excuse.

Some nights the best thing I can build is nothing. I close the laptop and go to bed.

Full disclosure: I wrote this article at 5 am before work. So yeah, perfect example. I'm working on it.

If you're feeling this too, you're not the only one. Take care of yourself out there.