在游戏中学习 AI

The history of artificial intelligence

From Turing (1950) to ChatGPT (2022): the key dates that shaped AI.

该内容暂无中文翻译，当前以英文显示。 阅读中文完整指南 →
本页目录

1. The origins (1943-1956)

Before the term "AI" even existed

1943: Warren McCulloch and Walter Pitts publish the first mathematical model of the artificial neuron. 1950: Alan Turing proposes, in "Computing Machinery and Intelligence", the test that bears his name. 1955-1956: John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon organize the Dartmouth workshop — McCarthy popularizes the term "artificial intelligence" there.

  1. Put these 3 founding milestones in chronological order

  2. Who popularized the term "artificial intelligence" at the Dartmouth workshop?

2. Early hopes and the first winter (1956-1980)

Success followed by disillusion

1958: Frank Rosenblatt's perceptron. 1966: ELIZA, Joseph Weizenbaum's first "chatbot". 1969: the book "Perceptrons" (Minsky & Papert) reveals the limits of single-layer networks. 1973-1974: the Lighthill report triggers the first "AI winter" — a funding collapse after unmet promises.

  1. What is an "AI winter"?

  2. ELIZA (1966) could genuinely understand human emotions.

3. The return of neural networks (1980-2000)

New tools, a second winter

1982: Hopfield networks (2024 Nobel Prize in Physics). 1986: backpropagation popularized by Rumelhart, Hinton and Williams. Late 1980s: the second AI winter, with the collapse of the expert-systems market. May 1997: Deep Blue (IBM) beats Garry Kasparov at chess.

  1. Put these events in chronological order

  2. Did Deep Blue (1997) beat Kasparov through learning like today's AI?

4. The deep learning revolution (2009-2016)

Three ingredients come together

2009: ImageNet (Fei-Fei Li), a huge labeled image database. 2012: AlexNet wins the ImageNet competition (15.3% error vs. 25.8% for 2nd place), trained on 2 graphics cards — the deep-learning turning point. 2013-2014: word2vec (Mikolov) then GANs (Goodfellow). March 2016: AlphaGo beats Lee Sedol 4-1 at Go, in Seoul.

  1. Put these 4 events in chronological order

  2. By how much did AlexNet reduce the error rate compared to the runner-up in the 2012 ImageNet competition?

5. From Transformers to today (2017-2026)

From architecture to mass adoption

June 12, 2017: "Attention Is All You Need", the Transformer architecture. May 28, 2020: GPT-3 (175 billion parameters). November 30, 2022: ChatGPT launches — 1 million users in 5 days. August 1, 2024: the EU AI Act enters into force. January 20, 2025: DeepSeek-R1, a Chinese open-weights model.

要点

Less than 6 years passed between the Transformer (2017) and ChatGPT (2022) — AI's pace keeps accelerating.

  1. Put these 4 recent milestones in chronological order

  2. How long after launch did ChatGPT pass one million users?

Boss challenge: the AI timeline

Sort each event by whether it belongs to AI's early history (before 2010) or the modern era (2010 onward).