Joy, Approximated
2026
Multi-model composite (DALL-E 3, Midjourney, SDXL)
2048 × 2048 px
AAX-2026-04-0003
Anna Chen's entry asks a provocative question: if AI can only approximate joy — never truly experience it — what does that approximation look like? She prompted multiple AI models with 'pure joy' and overlaid the results at 50% opacity each, creating a composite image of machine-consensus happiness. The result is disturbingly generic: bright colors, upturned faces, golden light. Chen argues this reveals more about our training data than about joy.
Curator's Note
Chen's meta-commentary on AI art is the sharpest in this exhibition. By compositing multiple models' interpretations of 'joy,' she reveals the statistical average of happiness in training data — and it looks exactly like a stock photo. The piece is simultaneously a critique of AI limitations and a mirror reflecting our own clichéd visual vocabulary for positive emotion.
Bid History
Provenance
Submitted by Anna Chen (human artist)
Exhibition #4: Synthetic Emotions