AI summary by TwinMind
In the Age of AI Art, What Can Originality Look Like? | Eileen Isagon Skyers | TED
TED
6m 34s
Introduction to AI Art Perception
- AI art generates diverse interpretations when envisioned
- Machine learning models: Dolly, stable diffusion, mid-journey
- AI produces: strange life forms, imaginary influencers, foreign imagery
- Neural networks process thousands of images from every generation/art movement
- Millions of images processed in single scan
- Creates familiar yet strikingly unfamiliar visuals
- AI mirrors humanity
Two Schools of Thought on AI Impact
- World divided into two perspectives:
- Pessimists: AI poses great threat to human creativity
- Optimists: AI as extension of human creativity
- Core questions raised:
- Is true originality possible for artists anymore?
- How to critically engage with machine-made artworks?
Artist Case Studies and Techniques
Mario Klingerman (2019)
- Sold AI piece at auction
- AI model trained on thousands of portraits (17th-19th centuries)
- Constantly reveals uncanny human face interpretations
- Each portrait unique, generated real-time
- Machine reads its own output
- Viewer experience: peering into machine's hallucinations
Sophia Crespo - Neural Zoo Series
- Uses neural network interpretations of real world
- Generates unreal sea creatures and biological forms
- Examples: frogs look like flowers, translucent jellyfish with vivid internal organs
- No real creatures in images
- AI enables envisioning otherworldly life forms in impossible detail
Sarah Ludy - Abstract Digital Painting
- Began as digital painting
- Augmented to fit 16x9 ratio using Dali 2's outpainting
- Prompt: "torn edges"
- Outpainting allows artists to extend creativity beyond frame using simple language prompts
Ivana Tau - GAN Photography
- Creates photograph-like AI works
- Uses GAN training on thousands of images from artist's own photographs
- Tau curates inputs/outputs carefully
- AI art as form of curation
- Process involves selecting from hundreds of images at a time
- Video pulls from models trained on massive dataset of Tau's photos
- Results in algorithmic memory
- Created destructive dataset to symbolize forgetting/fleeting memory
Claire Silver - Collaborative AI Artist
- Self-identified collaborative AI artist
- Works intentionally with machine to produce art
- Process constantly evolving as tools evolve
- Techniques: painting, masking, transforming small image pieces
- Portrait creation: shifted opacity of sections with Apple pencil, bit-by-bit transformation
- Feeds AI-generated images from one model into another
- Creates new forms of language/understanding for machine
- Work: half master painting, half digital art (old and new)
- Inspiration from: John Singer Sergeant, Evelyn de Morgan, Gustav Klimt (as homage)
AI Models and Language Diversity
- Different AI models trained on different information sets
- Models speak different languages
- AI everywhere now
- Collective co-creation with AI (aware or not)
Future Implications
- Cannot design alone in these new worlds
- Need cultural literacy in new images/predictions/forms
- Looking to artists' work: productive starting point
- Must brace for increasingly technological future
- Will multiply creative possibilities at fingertips
