The Modern Trolley Problem - Responsible AI Principals
summary
Demonstrates training an object‑detection model with an Azure Percept on a Raspberry Pi, using the bare minimum 15‑image minimum (and noting 100 images is more reasonable), then deploying it to an obstacle‑avoidance robot that learns cones and Lego figures; the live run avoids cones but mislabels a toy lion and targets it with only ~60% confidence, and source control even contains a commented‑out kill command, highlighting gaps in data, testing and accountability.
AI-written summary · about
chapters (14)
- 0:00Introduction and welcome
- 1:14Trolley problem overview
- 2:52Ethical frameworks
- 8:35Autonomous vehicles
- 14:02AI responsibility
- 19:33Facial recognition issues
- 22:01Responsible AI principles
- 24:41Safety and robot laws
- 27:12Computer vision applications
- 32:06Model training demonstration
- 40:51Privacy and security
- 44:01Transparency in AI
- 47:28Accountability in development
- 52:02Q&A and concluding thoughts
related talks
58mThe Modern Trolley Problem - Responsible AI Principles
01h 04mThe Modern Trolley Problem - Responsible AI Principles
01h 00mThe Modern Trolley Problem - Responsible AI Principles
04mRandom Interviews 3
13mEnhancing Object Detection & Avoidance Systems for Robots in Challenging Environments Using AI
15mRobot Hand Driven by Computer Vision and Reinforcement Learning
01h 03mThe Ethical Dilemma of AI: Save the World or Forsake It?
16mOn the ethical challenges raised by robots powered by AI