Skip to content
SHUMPEI KISHI
Go back

The Efficiency Paradox

I’ll be honest. I’m having a blast.

I’ve been going all-in on AI agents, automating everything I can get my hands on. After my day job, I dive straight into personal projects — and these days, that mostly means giving instructions to AI and reviewing what comes back. My Mac mini runs 24/7 as a home lab server. I half-expect it to melt.

And somehow, I’ve never been busier.

The Paradox

Here’s what nobody tells you about using AI to get things done faster: the number of decisions you make goes up, not down.

AI finishes tasks in seconds. So now I’m reviewing, validating, and course-correcting at a pace I never had to before. Then I study to get better at using AI. Then I study to verify whether AI’s outputs are actually correct.

I’m not sure that’s what “efficiency” was supposed to mean.

It reminds me of what happened when email replaced letters. Faster, easier, more efficient — and yet somehow everyone got more overwhelmed, not less. Every time technology says “you can do more,” humans respond with “okay, let’s do more.” We can’t help it.

The difference now is the speed is incomparable. Anthropic and Google are shipping new capabilities almost daily. No matter how much I learn, there’s always more to catch up on. At some point, it stops being an information problem and starts being a speed-of-the-era problem.

The Effort of Being Lazy

Here’s a truth about myself: I am fundamentally lazy.

I automate things because I want to do less. But when the automation breaks, I get frustrated and spend hours fixing it. I catch myself working on my phone. I’m always connected.

There’s a certain irony in how hard I work at trying not to work.

I Finally Understand Managers

Working with AI has given me an unexpected insight: I now understand what it feels like to manage a high-performing team.

When someone else — or something else — executes faster than you can, your job shifts. You’re no longer doing the work. You’re making judgment calls, setting direction, catching errors. The thinking doesn’t decrease. It increases.

I used to underestimate how much mental energy goes into directing others rather than doing the work yourself. Now I get it.

Knowing When to Stop

In the middle of all this, I’ve been making one deliberate choice: protecting time with my kids.

My sons are six and four. They don’t care about my automation pipelines. They want to play Minecraft together or drag me into their imaginary worlds. The other day, I caught them building an automated chicken cooker in Minecraft. I have no idea where they picked that up.

Every person I admire has told me the same thing: guard your family time fiercely. It always comes back to that.

Oh, and who reminded me to slow down and disconnect?

AI did. Of course it did.


I genuinely think this is one of the most exciting times to be alive. But excitement without intention is just noise. Efficiency without purpose is just speed.

So I keep automating. And I keep stepping away.

Both feel necessary.


Share this post on:

Previous Post
Localization Isn't Translation
Next Post
CMS Daishugo — Will Content Management Systems Survive the AI Era?