Machine Learning Street Talk The History of AI
9 snips
Feb 2, 2026 A brisk tour of AI's origins from philosophical questions about machine thought to the 1956 coinage that launched the field. Short-lived symbolic successes, AI winters, and the expert systems boom get highlighted. The narrative shifts to machine learning, the deep learning rise fueled by data and GPUs, and why today’s scaling creates new economic value and more accessible models ahead.
AI Snips
Chapters
Transcript
Episode notes
Thinking As Math Sparked Early AI
- The idea that thinking can be reduced to math and logic birthed early AI and the field's name in 1956.
- Jaeden Schafer explains researchers believed vision, language, and reasoning were near-solved and predicted non-human intelligence in ~20 years.
Symbolic Rules Worked Until The Real World Broke Them
- Symbolic AI used hand-coded if-then rules and succeeded only in narrow domains like chess and logic puzzles.
- Jaeden recounts how these brittle systems collapsed when faced with language ambiguity and noisy vision, leading to an AI winter and funding cuts.
Expert Systems Brought Money Then Disappointment
- Expert systems in the 1980s encoded specialist decision rules for doctors and engineers and attracted heavy corporate investment.
- Jaeden notes they were expensive, brittle to change, hard to maintain, and triggered another phase of disappointment when they failed to scale.