1955Aug 31, 1955
Foundations
A field gets its name.
A proposal for a summer research project put “artificial intelligence” on the page—and set an ambitious agenda for 1956.
THE HISTORY OF AI
A remarkable story.
Still being written.
The moments that moved AI forward.
What happened. Why it mattered. What came next.
26 momentsReset filters ↺
1997May 11, 1997
Foundations
A machine meets the world champion.
IBM’s Deep Blue defeated Garry Kasparov in a six-game chess match, a striking demonstration of specialized machine capability.
20062006
Foundations
More power for a different kind of problem.
NVIDIA unveiled CUDA, opening GPU parallel computing to a wider range of scientific and research workloads.
20122012
Research
Learning to see takes a leap.
AlexNet showed how a deep neural network trained with GPUs could dramatically improve large-scale image classification.
2014Jun 10, 2014
Research
Two networks learn from each other.
Generative adversarial networks introduced a training framework in which a generator and a discriminator improve through competition.
2016Mar 15, 2016
Research
A new move in an ancient game.
AlphaGo’s 4–1 match victory over Lee Sedol demonstrated the power of combining learning with search.
2017Jun 12, 2017
Research
A new way to pay attention.
The Transformer introduced a different architecture for working with language. A foundation for much of the AI that followed.
2018Oct 11, 2018
Research
Context becomes part of the meaning.
Google’s BERT research used bidirectional pre-training to learn language representations from surrounding context.
2020May 28, 2020
Research
A few examples can change the task.
The GPT-3 paper explored how a large language model could perform many tasks from instructions and examples in its input.
2020Nov 30, 2020
Research
AI helps explore the shape of life.
DeepMind announced AlphaFold’s breakthrough results on the protein-folding challenge at CASP14.
2021Jan 5, 2021
Creative AI
Describe an image. Imagine the possibilities.
DALL·E explored generating images from written descriptions, combining language with visual creation.
2021Jun 29, 2021
Building
Your editor gets a collaborator.
GitHub introduced Copilot as an AI pair programmer, bringing code suggestions directly into a developer’s workflow.
2022Aug 22, 2022
Creative AI
Image generation moves into more hands.
Stable Diffusion’s public release made model weights and code available for a broader range of image-making experiments.
2022Nov 30, 2022
Everyday AI · EDITOR’S PICK
AI becomes a conversation.
ChatGPT’s public research preview gave people a simple way to explore language AI: ask a question, then keep talking.
2023Feb 24, 2023
Research
More researchers can study large language models.
Meta introduced LLaMA, a family of foundation models intended to broaden research access to large language models.
2023Mar 14, 2023
Everyday AI
Another approach to a helpful assistant.
Anthropic introduced Claude more broadly through chat and an API, following testing with early partners.
2023Mar 14, 2023
Research
The conversation gets more capable.
GPT-4 introduced a large multimodal model that could accept image and text inputs and produce text outputs.
2023Dec 6, 2023
Research
Different kinds of information come together.
Google introduced Gemini, a family of models designed around multimodal capabilities.
2024May 13, 2024
Everyday AI
The interface becomes more natural.
OpenAI introduced GPT-4o with an emphasis on working across text, vision, and audio.
2024Jun 20, 2024
Building
From an answer to something you can use.
Claude’s Artifacts preview introduced a dedicated space to see and iterate on content alongside a conversation.
2024Nov 25, 2024
Agents · EDITOR’S PICK
AI starts connecting the dots.
The Model Context Protocol proposed an open standard for connecting AI assistants to data and tools.
2025Jan 22, 2025
Research
Reasoning becomes a research focus.
The DeepSeek-R1 paper described reinforcement-learning approaches for developing reasoning behavior in language models.
2025May 16, 2025
Building
A coding assistant takes on a task.
OpenAI introduced Codex as a cloud software-engineering agent that could work on tasks in isolated environments.
2025Jul 17, 2025
Agents · EDITOR’S PICK
From talking about work to doing it.
ChatGPT agent brought a virtual computer into the workflow, combining research and actions under user direction.
2026Jun 9, 2026
Agents
Capability and deployment evolve together.
Anthropic announced Claude Fable 5 and Mythos 5, pairing new capabilities with different access safeguards.
2026Sep 3, 2026
Agents
Longer tasks become part of the conversation.
GPT-6 Astra brought a new focus on computer use, software engineering, and complex professional workflows.
A curated collection, not an exhaustive history. Event dates and original references are listed on every story.
THE FUTURE HAS ROOM FOR YOU.
What if you tried?
One question. One idea. One small beginning.
Let’s find your first step