> There’s an obvious irony here: I use AI to build an app about AI energy consumption. But the alternative is ignorance, which seems worse.

I looked for this paragraph to be honest. AI energy consumption ignorance is worse than what exactly?

I think the appropriate takeaway here (the abstract notion that AI is energy consuming, but the numbers depend on who where what is analyzed) is to weigh more if an energy-expensive task should still be done to solve today’s problem. I’d rather be ignorant about the numbers but pick my problems right. I’m one of those people who are disappointed that some of the costs humanity is paying for AI has inevitably contributed to generation of copyright-violating nsfw unfunny longform videos, likewise a harmless and probably instructive green-themed project.

>day after day, for an entire year.

That’s the answer. The creator, other creators they have interacted with and their friends, they become your (para) friends too. Almost like comparing big-budget vacation with a random person vs hanging with people you dear.

Fair, but people will watch a random youtuber doing a video essay for the first time on an old topic like sbmm or AI in games and enjoy it yet find a 500 million dollar movie boring. I feel that your answer it correct, but there is something deeper than just para friendships to explain these other cases and I'm guilty of it too (watching a random youtubers video essay).

I see where you're coming from. Interested in verifying indeed it's something deeper. Leaning towards that it isn't, like there's nothing deep in debating sweet food vs salty food; sometimes you crave one and despise the other but not 7 days a week.

If you're the author/core maintainer of the codebase, well the norm is to ignore the PR entirely. But thank them for the improvement idea/bug report.

If you're just one of the people with commit access, probably your regression testing setup is due.

I think it's ok to go the "buildable" path because what's now built was once a hypothesis that the idea was feasible.

>what problem do people around you complain about without you bringing it up?

Their incompetent boss, lying politicians.

This, most problems are caused by other people's poor behavior, and the only way AI can fix that is through being able to impose consequences on them - and I don't imagine that startup will get funded

I presume this kind of problems is almost mathematically non-existent until your tool's package registry grow a certain size. curious to learn how true that is.