Freakonomics Radio 619. How to Poison the A.I. Machine
457 snips
Jan 24, 2025 Ben Zhao, a computer science professor at the University of Chicago, focuses on adversarial machine learning and safeguards against AI-related harms. He unveils tools like Glaze and Nightshade designed to protect artists from A.I. misuse and shares strategies to counter surveillance tech. Zhao warns of the A.I. bubble's burst, examines the ethical dilemmas within tech, and discusses the tension between innovation and protecting intellectual property in the creative realm. His insights draw a vivid picture of the evolving landscape of AI and its impact on art.
AI Snips
Chapters
Books
Transcript
Episode notes
AI Art Forgery
- Artist Kim Van Doen noticed AI-generated knockoffs of her work online.
- Her original artwork had been scraped from the internet without permission to train AI models.
Ineffective Opt-Outs
- Content creator opt-out programs from AI companies rely on their good faith.
- Proving unauthorized AI training is technologically challenging, making opt-outs ineffective.
AI: Identity Theft for Artists
- AI art generators can cause harm through identity theft by mimicking an artist’s unique style.
- Decades of artists’ skill-honing can be stolen, impacting their livelihood.
