I have some exposure to utility regulation and from what I can tell some of the AI companies are "good actors" and willing to shoulder some of the burden. But others are pretty adversarial and want a free lunch.

I can see from a lot of replies the "cool" threshold is undefined, but here goes:

For myself it let me finish a project I started a year ago for measuring how much home energy efficiency upgrades will reduce my AC usage. I bought a pile of Raspberry Pi Picos and turned them mostly into temperature reading devices, but also one that can detect when my AC turns on.

So I can record how often my AC runs and I can record the temperature at various points around the house, which lets me compare like-for-like before-and-after.

The easy but unrealistic way to accomplish what I want is to use Python. It gives me access to a file system, a shell, and all sorts of other niceties. But I wanted to run these on two AA batteries and based upon my measurements they would last about 2 weeks. I tested using C instead and they should last 4 months. That's long enough for my use case. There's enough flash storage for that time period too.

However this means I need to write all the utilities for configuring the Picos myself. There's all sorts of annoying things such as having to set the clock (picos lose it anytime they lose power), having to write directly to flash memory (no operating system), having to write a utility for exporting that data from flash memory, and so on.

And AI coding let me burn through a pile of code I knew how to write but didn't care to spend my weekends doing so.

The pattern is the same for my friends who are software devs. And yeah, you're probably never going to see any of it, but that's not why they're making it, they don't want the maintenance burden.

When our fridge stopped fridging, we got it fixed for $300: this included replacing the compressor and the coils. When our dishwasher stopped washing, we paid $250 to have 3 or so things fixed at once. And so on.

I don't know if any appliance makers offer this, but if LG still offers it when we eventually replace, they're going to be on the top of my list.

The type of code you're writing isn't special, it's the way people have written lots of clojure programs for over a decade.

Also, I think you'll find that taking care of someone who can't take care of themselves is a lot of work. I had to do it for my mom for 6 months and its a ton of stuff. Talking to doctors. Arranging appointments. Etc.

The southwest is basically exporting its water very cheaply in the form of agriculture. Why when its such a constrained resource here?

https://claude.ai/share/abb3e667-252a-4b34-86f7-a064ba260d2a

This reminds me of something funny I noticed about AI. Let's say you ask it what it thinks of an email you just drafted. It will provide corrections.

Delete that session, and ask it about the corrected email. It will provide more corrections.

Repeat. It always provides more corrections. Sometimes returning the recommended email back to a previous state.

This is basically what's gonna happen when people argue-from-AI. It's the same cycle but because control is distributed the individuals participating can't see how stupidly pointless it is.

Speed is seductive.

The bar isn't "this is a known good contributor". Its "this is a known good contributor working in a space they have knowledge in and has a track record of actually checking and thinking about LLM output before submitting it." It's much higher and I don't see how you can approve people on an organization-wide basis.