On Wisdom 55: Wise of the Machines (with Sina Fazelpour)
Aug 5, 2023
Sina Fazelpour, an assistant professor of philosophy and computer science at Northeastern University, explores the intersection of AI, ethics, and societal impact. They delve into the challenges of bias in algorithms and emphasize the importance of diversity in AI systems. Sina discusses how wisdom can be seen as an awareness of one’s social position and the need for humility in decision-making. The conversation highlights the potential of AI to enhance human creativity and problem-solving while cautioning against simplistic approaches to algorithmic alignment.
AI Snips
Chapters
Books
Transcript
Episode notes
Make Uncertainty Incentivized
- Intellectual humility requires both exposure to diverse perspectives and incentives that allow admitting uncertainty.
- Organizational incentives often punish saying 'I don't know,' so changing incentives matters more than just teaching humility.
Bias Is A Value Specification Problem
- 'Bias' means deviation from a chosen norm, so value specification is the foundational challenge.
- Diagnosing bias requires plural interventions across data, models, tasks, and organizational use.
Match Fixes To Bias Causes
- Diagnose the source of bias before choosing interventions; different causes need different fixes.
- Intervene at data, model, organizational, or regulatory levels depending on where bias arises.
