Potentially nefarious, but certainly ambiguous patterns that impact user trust with questionable user value
Inform users about shortcomings or risks in the model or the technology overall
Control how the model remembers and uses your data
Signal expectation gaps or errors in the model – but is that clear to the user?
