The AI snake is already eating its own tail. xAI’s Grok responds with ChatGPT error messages because there’s now so much ChatGPT content on the web that new LLMs are being trained on LLM output.
One can imagine LLMs getting worse over time in much the same way Google Search has gotten worse over time due to SEO spam.
Follow
@carnage4life At least as long as they indiscriminately trawl for data, yeah, that seems inevitable.
I'm hoping this pushes us faster into better-curated datasets, both for quality and consent.
