The worm is starting to turn....

Sure is looking like all the AI tools are bumping up against something.  Whether it’s adoption, hallucinations or productiziation there will be challenges to bring these tools to bear in the work and personal world. The idea that as more AI slop ends up in the data set, the more it goes off the rails is an interesting one. The other interesting concern is raised in this article around the concept of RAG (Retrieval-Augmented Generation) trouble.  RAG is an LLM accessing your own data sets and using them to produce better and less hallucinatory results.  While it improves it it comes with it’s own challenges and risks.  LLM management in the enterprise will be something that will have to be better understood and managed.

AI model collapse is not what we paid for • The Register: