Are Privacy and Intellectual Property Still Relevant? Perspectives From History, Biology, Math, & AI
summary
Proofs and examples land here: code that published every possible melody on GitHub and LibraryofBabel text generators challenge copyright; PCG random streams can index "Hello World"; diffusion models show 15-to-1 or 50,000-to-1 compression yet still memorize and reconstruct images; ChatGPT-class models show ~7:1 text compression; differential privacy and synthetic data preserve statistics while preventing reidentification; insurance telematics and university COVID testing produce perverse incentives; many systems spend roughly 20–30% on defense while privacy gets under 1%.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:13Copyright limits
- 4:25Ancient IP origins
- 6:14Patent and copyright evolution
- 8:01Human creation defaults
- 9:26Patent boundaries in math/biology
- 10:50AI and copyrightable content
- 12:57Myerson-Satterthwaite theorem
- 16:27IP enforcement vs innovation
- 19:36Game theory equilibria
- 22:43Security through obscurity
- 25:16Lanchester’s law in cyber
- 28:03Privacy’s historical roots
- 30:54Privacy leaks from data
- 34:24Why privacy matters
- 37:14Differential privacy basics
- 39:43AI’s impact on networks
- 40:45Open source AI governance
related talks
01h 27mLLM Privacy Paradox: Balancing Data Utility...
01h 09mAssumptions of Computer Science Interacting with Assumptions of Text-driven Normativities
47mUnderstanding the Legal Context of AI
30mPrivate Generative Models: Balancing Privacy and Utility in AI
01h 17mTutorial: Privacy, Part I
42mPennywise - Invisible privacy risk in the new AI era
29mKeynote Presentation: Minding the (Privacy Research/Practice) Gap
01h 43mPerspective