eieio.games

by nolen royalty

Will AI Pet My Dog for Me?

What work do I want to skip?

Jan 19, 2026

I have a dog. Her name is Gabby. She’s lovely.

Gabby and me sitting outside on a couch on a beautiful sunny summer day. Our noses are  touching and I'm smiling. Gabby is a yellow dog that weighs about 45 pounds.

Gabby and me last summer

She’s a lot of work. A dog walker handles Gabby’s afternoon walks. It’s nice to outsource the afternoon walk since it interrupts my day.

Gabby’s appetite for play and affection is insatiable. I sometimes want to outsource more of the work of taking care of her.

To align my goals of “building as much weird software as possible” and “taking care of my dog,” I could outsource all of the work of caring for Gabby. Someone else could walk her and feed her and pet her, leaving me free to find better UUIDs.

My time is valuable; this would be an efficient use of my money.

But I don’t do that. I like petting my dog.

I like to understand

I like to understand things. I like to share that understanding with others. This is my favorite part of building software. Until recently I had to understand the software that I developed.

LLMs can now type most of my code for me. And if I want, I can often accept that code without understanding how it works. That’s a big change.

LLMs have changed a lot of things! I am good at generating lots of code quickly; that skill is less valuable than it was 5 years ago. I’m pretty quick in vim; that’s probably not as important either.

It’s uncomfortable that some skills I’ve honed over years are becoming less valuable. But code has fussy syntax, boring parts, and boilerplate. It’s exciting to have agents that can take my code on its afternoon walk.

It’s more uncomfortable to be able to skip the understanding. For me, that’s petting the dog.

I can still choose to understand

Many of my blogs are about understanding useless computer problems. They would be bad posts if they said “I asked an LLM to solve this problem and it did.”

Screenshot of the claude web interface. I have typed 'Write an essay comparing 'using LLMs to code without understanding their outputs' to 'hiring someone to pet and play with my dog,' focusing on how my love of coding is tied to my love of understanding'

I did not do this

Nobody can make me use LLM outputs that I don’t understand for this blog.

But my blog’s style doesn’t control the software industry. Understanding can be a chore or expense on the way to shipping. Maybe it doesn’t make sense to pay to pet the dog anymore.

That may change the internet’s appetite for my explanations. Selfishly, I hope people still enjoy reading them. I hope that I still want to write them.

Humans like to understand

Understanding is fun even when it isn’t necessary.

Bartosz Ciechanowski’s blog is a wonder of the internet. It’s all illustrations of things that I don’t need to understand. Primitive Technology is an extremely popular youtube channel about building things in nature from scratch. My favorite video game is about understanding a new world.

The joy of understanding is a core human experience for me. If you read my blog, I bet it is for you too.

I don’t understand what comes next

Lots of LLM discussion is about whether programmers are about to be out of a job. I have nothing to contribute there. And if you’re talking about this blog on the internet, I’d like to nudge you away from a discussion you’ve probably had too many times in the past 3 years.

My immediate fear isn’t that I’ll be out of a job. My fear is that I’ll lose the part of my job that I love the most.

That fear may be wildly off base. Maybe LLM-assisted coding will just change what I need to understand and give me more time than ever to pet the dog. It’s hard to say.

But regardless, I’m confident that I’ll still love understanding for its own sake. I think you will too. That’s comforting to me.

Anyway. Here’s Gabby starting to understand how to rebound a balloon.

She's pretty good at this now.

Thanks for reading!

Keep up with me on my socials 👆

Or sub to my newsletter here! 👇