The Road to Artificial Superintelligence.
Trace the historical progression of AI from Turing's early mathematical foundations to the 2026 frontier of agentic reasoning models. Every milestone is linked directly to primary source papers and official documents.
Alan Turing Proposes the Turing Test
Alan Turing publishes his seminal paper 'Computing Machinery and Intelligence', introducing the 'imitation game' (now known as the Turing Test) to evaluate machine intelligence.
The Dartmouth Workshop
A summer research project organized by John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon at Dartmouth College, widely credited with coining the term 'Artificial Intelligence'.
Frank Rosenblatt Invents the Perceptron
Rosenblatt introduces the Perceptron, the earliest artificial neural network architecture designed for pattern recognition and binary classification.
Popularization of Backpropagation
Rumelhart, Hinton, and Williams publish in Nature, demonstrating how the backpropagation algorithm can train multi-layer neural networks to learn internal hidden representations.
Deep Blue Defeats Garry Kasparov
IBM's chess-playing computer Deep Blue defeats reigning world champion Garry Kasparov in a historic six-game match under tournament conditions.
AlexNet Wins the ImageNet Challenge
Alex Krizhevsky, Ilya Sutskever, and Geoff Hinton train a deep convolutional neural network (AlexNet) using GPU acceleration, achieving a massive win on ImageNet.
Ian Goodfellow Invents GANs
Goodfellow et al. introduce Generative Adversarial Networks (GANs), setting up a competitive game between a generator and a discriminator model.
ResNet: Deep Residual Learning introduced
Kaiming He and colleagues introduce Deep Residual Learning, using skip connections to train neural networks containing over 150 layers.
AlphaGo Defeats Grandmaster Lee Sedol
Google DeepMind's AlphaGo defeats 18-time world Go champion Lee Sedol 4-1 in a televised match in Seoul.
The Transformer Architecture Introduced
Google researchers publish 'Attention Is All You Need', introducing the Transformer architecture built entirely on self-attention mechanisms.
AlphaStar Reaches Grandmaster in StarCraft II
DeepMind's AlphaStar agent achieves grandmaster ranking in StarCraft II, mastering imperfect information and real-time strategy.
OpenAI Releases GPT-3
OpenAI publishes 'Language Models are Few-Shot Learners', introducing the 175-billion parameter GPT-3 model.
OpenAI Launches ChatGPT
OpenAI introduces ChatGPT, a conversational chatbot aligned via Reinforcement Learning from Human Feedback (RLHF).
GPT-4 and economic impact reviews
OpenAI releases GPT-4, demonstrating human-level performance on legal/medical exams. Goldman Sachs projects AI will raise global GDP by 7%.
Google Genie & Willow quantum chip
Google DeepMind introduces Genie (playable world models), and Quantum AI announces the Willow chip achieving quantum error correction below threshold.
Project Sid: 1,000 Autonomous Agents in Minecraft
Altera AI runs a simulation of 1,000 cooperative language model agents in Minecraft, showing trade networks and social structures.
The Rise of RL & Reasoning Models
The release of reasoning models (like DeepSeek-R1 and OpenAI's o-series) shifts the paradigm from pre-training compute to inference-time scaling.
State of AI: Enterprise adoption reaches 88%
McKinsey's annual survey confirms that generative AI adoption has surged to 88% of organizations using it in at least one business function.
OpenAI Launches Flagship GPT-5.5
OpenAI releases GPT-5.5, a highly optimized agentic model family designed for multi-step task execution, tool use, and self-correction.
Claude Fable 5 & coding benchmark breakthroughs
Anthropic launches Claude Fable 5, achieving an 80.3% solve rate on the SWE-bench Pro coding benchmark, and Claude Code reaches $2.5B ARR.
SpaceX Acquires xAI in Landmark Compute Deal
SpaceX formally acquires xAI for $20B to combine physical engineering, Starlink satellite backbones, Optimus humanoid brains, and sovereign defense networks.
Anthropic Releases Claude Sonnet 5 & Claude Science
Anthropic launches Claude Sonnet 5 for mainstream agentic workflows, alongside Claude Science, a dedicated browser-based workbench for scientific research.
Fable 5 Global Redeployment & Jailbreak Framework
After a temporary mid-June suspension due to government security concerns, Claude Fable 5 returns globally with a shared safety framework for scoring jailbreak severity.
Google Releases DiffusionGemma
Google open-sources DiffusionGemma, demonstrating parallel token drafting architectures that generate text up to 4x faster.
Moonshot AI Launches Kimi K2.7-Code MoE
Moonshot AI rolls out Kimi K2.7-Code, a 1-trillion parameter Mixture-of-Experts coding model engineered for long-context program synthesis.
Zhipu AI Announces GLM-5.2
Zhipu AI launches GLM-5.2, bringing advanced agentic planning and code execution parameters to the GLM open weights stack.
OpenAI Launches GPT-5.6 Series (Sol, Terra, Luna)
OpenAI introduces GPT-5.6 in a limited preview (Sol flagship, Terra balanced, Luna fast). Release is restricted to vetted partner organizations at the government's request.
Decart AI Unveils Oasis Generative Engine
Decart AI launches Oasis, the first streamed real-time generative world model, simulating Minecraft gameplay frame-by-frame directly from user input.
Boston Dynamics Unveils Production Electric Atlas
Boston Dynamics releases the fully electric version of its Atlas humanoid with 56 DoF, committing initial production to Hyundai and Google DeepMind.
Figure AI Deploys Figure 03 at BMW Group Plant
Figure AI deploys its Figure 03 humanoid robots at BMW's Spartanburg plant, executing complex component sequencing under the Helix 02 AI system.
UAE's G42 Launches Sovereign AI Greenshield
Abu Dhabi's G42 partners with Microsoft to launch Greenshield and Digital Embassies, guaranteeing localized data residency for agentic AI workloads.
US Issues NSPM-11 on National Security AI
The White House issues National Security Presidential Memorandum-11 to accelerate the acquisition of multi-vendor commercial frontier AI models.