This really matches up to my experience on long research projects with Claude.
It’s very hard to remove information - Claude has a habit of recording things all over the place and will happily treat things as facts even after they’ve been disproved.
What is currently true can get easily contaminated with old “facts”.
One of my rules of thumb is to minimize the amount of "negative knowledge" an LLM sees, as the concept introduced just hangs around in the context way too strong.
The mnemonic I use for that is "do not think of a polar bear".
Comments
This really matches up to my experience on long research projects with Claude.
It’s very hard to remove information - Claude has a habit of recording things all over the place and will happily treat things as facts even after they’ve been disproved.
What is currently true can get easily contaminated with old “facts”.
One of my rules of thumb is to minimize the amount of "negative knowledge" an LLM sees, as the concept introduced just hangs around in the context way too strong.
The mnemonic I use for that is "do not think of a polar bear".
https://en.wikipedia.org/wiki/Ironic_process_theory