A silhouette parlor’s CLIP-steered generator passes a “looks crisp” check but invents a sitter who never visited. Can the studio file that under a trustworthy model?
Select an answer to reveal the explanation.
Short Explanation
A sharp portrait of someone who never walked in is still fiction. Trustworthy content is about honesty of origin and subject—not how glossy the pixels look.
Full Explanation
A silhouette parlor’s CLIP-steered generator that passes a “looks crisp” check but invents a sitter who never visited cannot be filed under a trustworthy model on visual polish alone. No: visual polish does not equal content authenticity when the still invents a person who was not there. Domain 7 separates aesthetics from truthfulness for civic portrait lies.
Yes because crisp pixels mean trustworthy fails: crispness is not authenticity. Yes if CLIP embeddings score above an arbitrary threshold fails: similarity scores do not prove a real visit happened. Yes after converting the still to a Domain 5 gallery collage fails: collage presentation does not repair invented identity.
Exam caveat: trustworthy content authenticity ≠ perceptual quality metrics. Invented sitters are trust failures even when beautiful.
Operational check: require a visitor or archive match before filing portraits as authentic; quarantine invented sitters; disclose generated speculative portraits clearly in the municipal parlor. In municipal trustworthy-AI stewardship, record the disclosure, consent, or bias decision and the rejected shortcuts in the gallery change log so future stewards protect residents from undisclosed synthesis and stereotyped generation. Re-audit labels, consent records, and bias samples after the next model or exhibit update to keep the city's public trust intact.