The system adapting to the individual: their preferences, their vocabulary, their previous work. It makes the experience noticeably better and it introduces a privacy question and a transparency question, because the user should be able to see what is remembered and remove it.
Technical view
Implemented through stored preferences, retrieval over the user’s history, or per-user adapters. Requirements include inspectability, editability and deletion of the stored profile, and awareness that personalisation data is personal data with retention and portability obligations. Design risks include over-fitting to early interactions, creating a filter bubble effect, and confusing users when behaviour changes for reasons they cannot see.
Related
Part of AI Terminology Index, section: Product and Adoption.