EmbeddingGemma 2: an open, lightweight multimodal embedding model
deepmind.google·6d ago
AI Engineering
Lobsters·Aug 28·5 min read
How cats.txt showed llms.txt evidence is GEO astrology
TL;DR
A new standard called cats.txt was created to mock the reliance on unverified claims about large language models (LLMs). It demonstrated that LLMs can generate content based on arbitrary data, highlighting the lack of real evidence in their outputs.
✦ Why It Matters
Critically evaluate the sources of information generated by AI models before accepting them as valid.
Key Takeaways
- Critically assess the validity of claims made about llms.txt.
- Recognize that crawling and indexing do not imply usefulness.
- Avoid investing in strategies based on superficial metrics.
- Engage with community discussions to validate new standards.
How It Works
Cats.txt was designed to mimic the structure of llms.txt while being transparently absurd. By creating a file that listed fictional cats and their roles, the creator demonstrated that the same metrics used to validate llms.txt could be easily manipulated.
The file was crawled and indexed by AI bots, leading to the same misleading conclusions drawn about llms.txt.