Summary
Fireship traces the intellectual lineage of modern AI through ten of the most consequential computer science papers of the past century, connecting each directly to how today’s large language models work. The video opens with Alan Turing’s 1936 paper on computable numbers — which defined what an algorithm is while proving the halting problem is unsolvable — then moves to Claude Shannon’s 1948 information theory paper, which introduced the bit and the entropy concept that underlies modern loss functions.
Subsequent papers covered include Frank Rosenblatt’s perceptron (1958), the Minsky and Papert book that triggered the first AI winter, Leslie Lamport’s work on logical clocks for distributed systems (essential for large-scale GPU training), Rumelhart and Hinton’s backpropagation paper (1986), Yann LeCun’s MNIST work, and the 2012 AlexNet paper that ended the second AI winter. The video closes with the 2017 Google paper “Attention Is All You Need,” which introduced the transformer architecture underpinning GPT, Claude, and Gemini.
Fireship’s signature fast-paced style makes dense material accessible without sacrificing accuracy. The connective thread throughout is how each paper accidentally or deliberately contributed something the authors didn’t fully anticipate — Shannon wasn’t trying to build AI, but his entropy math became the foundation for token prediction. A strong primer for developers and enthusiasts who want to understand why the current generation of AI systems works the way it does.
📺 Source: Fireship · Published June 17, 2026
🏷️ Format: Deep Dive
