Post #2285641

@vantiss@social.treehouse.systems Yeah, I read that one and was confused, it didn't even argue what it sets out to argue from my interpretation, the title and description seems to talk about the LLM coding a better LLM that's coding a better LLM etc, while the article argues that feeding it's own output into its own training material makes it worse, which seems kind of self evident. I was more confused after having read it.