Post by @zzt@mas.to

https://blog.fallible.net/it-breaks-a-village/

this is an extremely important read, even if you don’t care about the Rust game engine Bevy.

anti-AI policies are broadly popular, and pro-AI policies destroy communities. this article is an excellent summary of a project going from an apparent (but not actual) anti-AI stance to a pro-AI one, on the false pretenses that AI boosters were subject to harassment and witch hunts. those false pretenses are already common, and you’re going to see them a lot more soon because this is a playbook.

View

3 Boosts

0 Likes

[object Object]

@zzt@mas.to 3 hr. ago

quoting one section that’s powerful even if you don’t care about Bevy or Rust, but the whole post is excellent:

> In software we have this strange position taking hold, that our work is as valueless as the output of an LLM. Not that LLM output is just as valuable, no, that we are of as little value as it. This sometimes gets communicated as "They're just as good as you!" which is a slight against the programmers being talked about, more than anything.
>
> Software is the outlier in how people culturally value LLM output. Game engines have a relationship to people outside of software.
>
> Nobody intentionally buys "AI assets" from online marketplaces. Nobody asks "AI artists" for commissions or collaboration. The people who follow "AI art" accounts are... "AI" boosters, for the most part. Some people who are caught up in the spectacle. "AI" works on steam instantly receive backlash, and are seeing an increasing amount of backlash over time.

[object Object]

@zzt@mas.to 3 hr. ago

> Game Developers see "AI" as a liability, Players see "AI" output as asset-flip vapourware. When people hear that "AI" is part of a project, many of them lose trust in that project. This holds true even when it's "just" code. It's viewed as an abdication of responsibility, an abdication of care.
>
> People are consistently picking up on "AI" tells in video, writing, images, no matter the "improvements" in models. This is a matter of needing to filter out information from the world. "AI output" is noise, not signal, and in order to make human connections over a mediation layer like the internet you need to be able to filter out noise quickly.

[object Object]

@zzt@mas.to 2 hr. ago

also, it’s important to point this out: Bevy went from an apparently anti-AI but in practice centrist policy to a policy that’s being presented as centrist but is in practice extremely pro-AI

if you wanted a worked example of why centrist AI policies are horrible, this is it

Marisa

@witchgirls@woof.tech 2 hr. ago

@zzt It was fated to happen after one of the main developers made a super mega long post talking about how yes LLMs are bad but they're innevitable and you can't resist them so just use them and hope the governments regulate them in time I can buy you some claude tokens if you ask nicely :)

1+ more replies