Harvard Data Science Review Podcast What Can We Learn From The Histories of AI: A Conversation With Stephanie Dick
10 snips
Apr 30, 2026 
Stephanie Dick, historian of mathematics and technology and assistant professor at Simon Fraser University, explores how past ideas about knowledge shape today’s AI. She traces three historical acts of AI and shows how data and design reflect human choices. She questions the single-minded pursuit of artificial general intelligence and argues for plural, embodied, and relational notions of intelligence.
AI Snips
Chapters
Books
Transcript
Episode notes
Study History Before Declaring AI Revolutionary
- Use history to ground AI innovation: study past technical-social choices to predict which interventions actually produce meaningful change.
- Historical patterns show technologies often reproduce existing power structures instead of transforming them.
Three Acts Explain How AI Keeps Changing
- AI evolved through three distinct paradigms: symbolic reasoning (1950s–60s), expert systems, and data-driven pattern recognition (machine learning/LLMs).
- Stephanie Dick ties each act to a different theory of intelligence, e.g., early AI modeled intelligence as formal rule-based reasoning used in chess and logic.
Raw Data Is Always Shaped By Human Values
- All data reflect human choices; there is no neutral "raw" data because collection priorities encode values and power structures.
- Example: early U.S. censuses prioritized white men aged 25–45, shaping what was counted as the nation's capacity.
