Mastodon Feed: Post

Mastodon Feed

cstanhope@social.coop ("Your weary 'net denizen") wrote:

I didn't want to learn this because I feel agentic development is worse than a waste of time. But in following up on a (now pulled) story from Ars Technica about the situation above where Ars apparently used made up quotes from the target of the "AI" harassment, I learned that the software running these agents have a SOUL.md document that feeds into the prompt that the software can modify on its own (generating bizarre feedback loops I am sure).

https://web.archive.org/web/20260214062635/https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

https://web.archive.org/web/20260214060018/https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-2/

This is all an incredible waste of time and energy. It's an attack on individuals, culture, and society. We should spend less time figuring out how people deployed these "agents" and more time simply saying "no".

I hope not, but maybe, as Shambaugh says, there's a quarter of developers that "side" with these pieces of software. I think that's all the more reason we must consistently say no before it gets worse.