TLDR InfoSec · MISCELLANEOUS
ASSUMPTIONS_MADE
Summary from the TLDR InfoSec newsletter
When LLMs and AI agents respond to user inputs, they inherently make certain assumptions to provide context and accurate responses to a user. These assumptions should be explicitly stated in an assumptions_made metadata field. This would provide more insight into improving LLMs and AI agents.