chapter 00 / history · prologue · estimated time 45-60 min

Seventy Years of AI
A Brief Rollercoaster History

AUDIO // Chapter Audio Guide

Artificial intelligence did not march in a straight line to where we are today. It lived through two fevered springs, two bone-chilling winters, several wars over methodology (symbolism vs. connectionism, rules vs. learning), and a handful of moments when "the entire field was rewritten by a single paper." Understanding this history is not just nostalgia—the cause of every winter (overpromising) and the recipe for every breakthrough (algorithm × data × compute arriving at the same time) are still playing out today.

Interactive Timeline: 1943 → Now

Below is a canvas you can pan and zoom without limit. Seven lanes correspond to AI's seven main threads, and the larger a dot is, the more important the event. Use the scroll wheel to zoom into the 2012-2026 stretch—and you'll get a visceral sense of what "more has happened in the last decade than in the previous seventy years combined" really means.

This chart contains about 55 milestones, and every node opens to a substantive technical reading (not a "Model X released in year Y" laundry list). It will keep expanding as the course evolves—the right edge of the timeline is open-ended, just like the field itself.

Act One (1943-1973): Birth and the First Spring

AI's starting point was not computer science but a marriage of neuroscience and logic: in 1943, McCulloch and Pitts proved that networks of neurons can implement arbitrary logical operations. In 1950, Turing gave an operational criterion for "what counts as intelligence." In 1956, at the Dartmouth workshop, "Artificial Intelligence" was formally named and established as a discipline.

Over the next twenty years, two routes sprouted at the same time but met very different fates: symbolism (logic, search, rules) became the mainstream orthodoxy—chess programs, theorem provers, ELIZA; connectionism (neural networks), after the perceptron frenzy of 1957, was cast into the cold by Minsky's 1969 book Perceptrons, which proved that "a single-layer network can't even learn XOR."

The direct cause of the first winter (roughly 1974-1980) was overpromising: researchers had promised governments general intelligence and fully automatic translation within a decade, yet failed to solve even the basic problems. In 1973 the UK's Lighthill Report categorically dismissed the value of AI research, and funding fell off a cliff. Remember this pattern—the "promise-disappointment-defunding" cycle has happened more than once in AI's history, and it remains the best sobering antidote when we discuss "when will AGI arrive" today.

Act Two (1980-1995): The Expert-System Bubble and the Connectionist Revival

In the 1980s, AI regained its commercial ticket through expert systems: encoding human experts' knowledge into thousands of if-then rules. XCON saved DEC tens of millions of dollars a year, Japan launched its "Fifth Generation Computer" program, and the US responded with matching investment—a second spring. But the maintenance cost of rule bases exploded with scale, the bubble burst after 1987, and the second winter set in.

The real spark was lit inside the cold cell: in 1986 backpropagation made multilayer networks trainable (a direct answer to the 1969 critique), in 1989 LeCun's CNN began reading handwritten checks, and in 1997 the LSTM solved the vanishing-gradient problem in recurrent networks. The connectionists laid the infrastructure brick by brick in years when no one was cheering—Hinton, LeCun, and Bengio would later share the Turing Award, a reward precisely for those years spent riding the bench.

Act Three (1995-2011): The Golden Age of Statistical Learning

Two winters taught the field humility: don't talk about "intelligence," solve concrete problems first. The SVM (1995) dominated academia with its elegant convex-optimization theory and powerful performance on small data; random forests and boosting engineered "the wisdom of the crowd"; machine learning quietly entered spam filtering, credit scoring, and search ranking—AI was everywhere, just nobody called it AI. The methodology distilled in this period (train/test splits, cross-validation, regularization) is exactly the fundamentals you just learned in Chapter 1.

Meanwhile, three seemingly unrelated events were storing up energy for the detonation: the internet accumulated massive amounts of data; Fei-Fei Li released ImageNet in 2009; and NVIDIA's gaming GPUs were found to be remarkably well suited to matrix operations (CUDA was released in 2007). Algorithm, data, compute—the three ingredients were about to be complete.

Act Four (2012-2016): The Deep Learning Revolution

On September 30, 2012, AlexNet cut the error rate in the ImageNet competition from 26.2% to 15.3%. The shock of that number lay in its cliff-like drop: previous champions had improved by a toothpaste-squeeze of 1-2 percentage points each year. Within three years, every subfield of computer vision was rewritten by CNNs; in 2015 ResNet surpassed human-level performance on ImageNet; and in 2016 AlphaGo's move 37 let the whole world—especially China and Korea—see machine "intuition" with their own eyes.

Why 2012? Every component in the AlexNet paper (CNN 1989, backpropagation 1986, GPU training had already been attempted by 2009) was nothing new. What was new is that they came together for the first time on enough data (ImageNet) and enough compute (two GTX 580s). This is a theme that recurs throughout this course: AI's leaps rarely come from brand-new theory, and more often from "an old idea that finally waited for the scale it needed."

Act Five (2017-2021): The Transformer and Scaling Laws

The 2017 paper Attention Is All You Need was meant to be just a machine-translation paper, yet it accidentally provided a general architecture that was a perfect fit for GPU parallelism and whose performance grew smoothly with scale. The GPT series began a high-stakes bet: scale up the same architecture tenfold, a hundredfold, generation after generation. In 2020, GPT-3 (175B) proved the bet right—scale brought not only quantitative change but the qualitative, emergent ability to "learn a new task from just a few examples." That same year, the Scaling Laws paper turned "buying intelligence with compute" into a computable engineering decision.

Act Six (2022-Now): The Era of LLMs and Agents

On November 30, 2022, ChatGPT launched. One hundred million users in two months. The pace that followed you've probably lived through yourself: GPT-4 and the move to closed source, LLaMA and the open-source ecosystem, multimodality, the reasoning models o1/R1 and the new dimension of "test-time compute," MCP and the tool ecosystem, agentic coding rewriting software engineering... The timeline gets crowded here, and that's not an illusion—it's the real acceleration of this field. Pull the timeline all the way to the right: the frontier questions of 2026—long-horizon agents, multi-agent collaboration, RL with verifiable rewards—are exactly the technical details that Chapters 8 and 11 of this course will dig into.

Having brought the history this far, this is the perfect moment to watch a documentary. DeepMind's official AlphaGo film documents in full the 2016 Seoul matches—it's both a visual archive of Act Four's climax and the best material for understanding the impact of "machine intuition":

VIDEO 01 · Documentary
AlphaGo - The Movie (the AlphaGo documentary)
Official DeepMind 1:30:27
Viewing Guide · Key Segments
  • 36:00 Game two, move 37, the "shoulder hit": the commentators' shock, Fan Hui's epiphany. AlphaGo estimated the probability a human would play this move at one in ten thousand—it was not imitating humans but had found, in self-play, a region of strategy space humans had never explored.
  • 1:10:00 Game four, Lee Sedol's move 78, the "divine move" that turned the tables—human creativity in a desperate situation is equally stunning, and AlphaGo's value network collapsed on the spot.
  • As you watch, ponder one question: how similar is AlphaGo's recipe (self-play + reinforcement learning + value network) to the recipe used today to train reasoning models (o1/R1)? Chapter 8 will give the answer.
  • We recommend turning on subtitles (Settings → Subtitles → Auto-translate → English).
What common structural cause underlay the two AI winters? Will today's LLM boom repeat the same mistakes?
Shared structure: (1) promises exceeded the ceiling of what the era's compute/data could support; (2) funders (governments/companies) expected "general intelligence," while deliverables were only narrow-domain tools; (3) a lack of quantifiable progress metrics meant that when confidence collapsed, the pessimistic narrative could not be falsified.

What's different today: LLMs already have real large-scale revenue (APIs, subscriptions, coding assistants), the technology has extrapolable engineering laws like Scaling Laws, and capability keeps advancing on measurable benchmarks—a "ground truth" that simply did not exist during the first two winters. What's the same: promises about the AGI timeline are once again running ahead of the evidence, and data-center capital expenditure is counted in the hundreds of billions of dollars. The conclusion is left to you: bubbles and revolutions have never been mutually exclusive—the dot-com bubble burst, yet the internet itself was real.

Chapter Quiz