Practical AI Computer scientists as rogue art historians
Apr 12, 2023
Amanda Wasielewski, an art historian at Uppsala University and author of "Computational Formalism, Art History and Machine Learning," reveals the fascinating overlap between art history and computer science. She discusses how computer scientists inadvertently resemble rogue art historians and how machine learning tools aid in research and art fraud detection. Key topics include the challenges of AI in classifying art, the ethical implications of generative technologies, and the impact of AI-generated visuals on societal trust and authenticity.
AI Snips
Chapters
Books
Transcript
Episode notes
Style in Art and AI
- In art history, "style" is complex and contested, unlike its simplistic use in AI.
- Generative AI's use of style markers like "impressionist" needs careful consideration.
Tank Classifier Story
- The "tank classifier" story highlights a machine learning challenge.
- It misclassified tanks based on weather, revealing a lack of true image understanding.
Art Category Nuances
- Inconsistent art categories can mislead machine learning findings.
- Terms like "action painting" and "abstract expressionism" have nuanced, contextual meanings.
