Self replicating genies - How to democratize and ensure ethics in AI
summary
Argument: current ML spawns algorithms that act like "self‑replicating genies" and require ethics, ownership and transparency. Demonstrations include Microsoft Face API (detects up to 22 faces, age/gender/expressions), Social Mapper (links one photo across social networks), and on‑device inference via TensorFlow.js for browser ML (teachable keyboard, pose/dance demos). Claims: bias appears when training sets are narrow, cloud vs on‑device changes privacy, and practical tools already exist to build useful assistive features and privacy‑preserving workflows.
AI-written summary · about
chapters (14)
- 0:00Start
- 4:09Reminders of genie in the bottle
- 5:46Let's start with some predictions
- 9:24Social Credit System
- 11:22Big brother is redundant...
- 13:15Leaving invisible marks...
- 14:07Artificial Intelligence Myths
- 18:41Machine learning helps us in a few
- 19:11Recommendation
- 21:52Classification
- 28:34Automated face mapping
- 33:13Intelligent responsive systems
- 34:25Our responsibilities
- 35:34Who controls our data? Who benefits?
related talks
16mMachine Intelligence With Humans in Mind
39mResponsible AI
26m(Un)ethical Artificial Intelligence: How to Keep the AI Fair for Everyone
43mAI Engineer Ethics: Shaping Responsible Generative AI
49mAn AI with an Agenda
47mEthics & AI: Identifying & preventing bias in predictive models
47mRoll for Initiative: how to make the world of AI a more ethical place
30mHow to Lie with AI: Understanding Bias, Ethics, and the Hidden Risks in ML