Breakthrough Timeline

History does not arrive labeled. When researchers invented backpropagation in 1986, most of the world had no idea anything significant had happened. When a neural network quietly won an image recognition contest in 2012, it was front-page news in AI research circles and almost invisible everywhere else. Only in retrospect do we see the shape of a story — how one development made the next one possible, how a single algorithm set off a chain of consequences that took a decade to fully arrive. In this lesson, you are going to do something historians do: build a timeline. But not just a list of dates and names — an annotated, connected timeline that shows the cause-and-effect relationships between breakthroughs. When you finish, you will have a working map of how AI got to where it is today.

The Breakthrough Reference Set

Here is a reference set of milestones to anchor your timeline. These are the events your annotations will connect. 1950 — Alan Turing publishes 'Computing Machinery and Intelligence,' introducing the question of whether machines can think and proposing what became known as the Turing Test. 1956 — The Dartmouth Workshop formally establishes AI as a field of research. 1986 — Backpropagation is published by Rumelhart, Hinton, and Williams, making it practical to train neural networks with many layers. 1997 — Deep Blue (IBM) defeats world chess champion Garry Kasparov — the first time an AI beats a reigning world champion under standard match conditions. 2006 — Geoffrey Hinton publishes work showing that deep networks can be initialized effectively, reigniting serious interest in deep learning. 2012 — AlexNet wins the ImageNet competition by a margin that shocks the research community and launches the modern deep learning era. 2016 — AlphaGo (DeepMind) defeats Go world champion Lee Sedol — Go was considered far too complex for computers to master for decades. 2017 — The transformer architecture is introduced in 'Attention Is All You Need,' becoming the foundation of modern language AI. 2018 — BERT and the first GPT model demonstrate that transformers scale powerfully when pre-trained on large text corpora. 2020 — GPT-3 demonstrates emergent capabilities including code generation, essay writing, and question answering from a single general-purpose model. 2020 — AlphaFold solves the protein folding problem, predicting 3D protein structures with experimental accuracy. 2022 — ChatGPT launches and reaches 100 million users in two months, making large language models a mainstream public technology. 2024 — Geoffrey Hinton and John Hopfield receive the Nobel Prize in Physics for foundational work on neural networks.

Cause and Effect: How Breakthroughs Build on Each Other

A timeline is more powerful when it shows connections, not just sequence. Consider a few chains: Backpropagation (1986) made it mathematically possible to train deep networks. But without GPUs fast enough to run the training and datasets large enough to generalize, it remained mostly theoretical. When both of those arrived in the early 2010s, AlexNet's victory in 2012 became possible. AlexNet demonstrated that deep learning worked at scale in computer vision. The excitement it generated pulled enormous talent, funding, and research energy into deep learning — accelerating progress across all AI subfields, not just vision. The attention mechanism that enabled the transformer (2017) grew partly from work on machine translation that was itself a product of deep learning. The transformer then became the architecture that made large language models like GPT-3 and ChatGPT possible. Each breakthrough is a stepping stone, not an island. Your timeline should show these steps.

What Your Timeline Reveals

When you step back and look at the completed timeline, a few patterns tend to emerge. Progress accelerates: the gap between major milestones has shortened over time. The 1956 Dartmouth Workshop and the 1986 backpropagation paper are thirty years apart. AlexNet (2012), the transformer (2017), GPT-3 (2020), and ChatGPT (2022) are all within a decade of each other. Progress is also cumulative: almost no breakthrough arrives without standing on the work of predecessors. And progress is unpredictable in its specific shape: most researchers in 1990 did not predict that a game of Go or the ability to write a persuasive essay would be early milestones of advanced AI capability. The timeline is not finished. It will continue to grow. You are living through entries that will eventually be added to it — and some of you will contribute directly to making those entries happen.