Last Week in AI More Ethical Quagmires, More Surveillance, and more Academia Talk
May 31, 2020
Delve into the ethical dilemmas of algorithms predicting criminal behavior and the historical biases that shape them. Join a heated discussion on the controversies surrounding facial recognition technology and the urgent need for regulations. Learn about the importance of broader impact statements in research to foster responsible scientific practices. Address the growing public skepticism towards self-driving cars and AI's involvement in surveillance, emphasizing challenges faced by researchers amid industry pressures.
AI Snips
Chapters
Transcript
Episode notes
AI Criminality Prediction Flawed
- AI can't predict criminality from faces; studies claiming high accuracy likely have flawed methodology.
- Such research has a dark past rooted in phrenology and promotes harmful biases.
Be Skeptical of AI Criminality Studies
- Be wary of AI research claiming high accuracy in predicting criminality from faces, often due to methodological issues.
- Consider the potential harms of research and be skeptical of studies promoting problematic biases.
Mistrust of Self-Driving Cars
- American mistrust of self-driving cars stems largely from lack of firsthand experience, not just reported incidents.
- Increased exposure and testing could boost public confidence in the technology.