Definition Sliding in Survey Measurement: The Chameleon-Like Problem of Counting Rainbow Colours in the Age of AI, Physical AI, the Metaverse and Bodily Augmentation

Identically worded survey questions and statistical classifications may cease to refer to equivalent objects when their referents and boundaries change across time or respondents. This is an established concern in survey methodology, response-shift research and measurement invariance. This research note considers its renewed urgency when artificial intelligence (AI), physical AI, virtual social spaces and bodily augmentation introduce new entities into familiar categories such as friend, competence, employment and authorship; the note calls this cross-domain problem Definition Sliding. Preserving a question may conceal changes in respondents’ referents, whereas redefining it may sever comparability with earlier answers. Where the referents of past answers were never recorded, their composition may be unrecoverable. As the number of affected domains, rate of change and combinations of classification boundaries increase, universal follow-up probing may be neither feasible nor neutral. The note proposes early parallel recording of unrestricted and referent-specific questions, with histories of classification changes.