A BRIEF HISTORY OF AI
1986
1980
1973
1958
1956
Early Criticism
The AI Comeback
Foundations
Link
Link
Link
Link
Link
Frank Rosenblatt's perceptron demonstrated a machine could learn simple patterns, inspiring early neural network research.
Expert systems capable of solving complex problems (like DEC's XCON rule-based program) delivered practical value in business, renewing interest in AI.
Researchers began asking if machines could "think." A group of scientists gathered at Dartmouth to research "machine learning" and coined the term "artificial intelligence."
The controversial Lighthill Report criticized AI research, stating that while AI could solve simple problems, it was not scalable to solve more complex real-world tasks. This led to a dramatic decrease in AI research in the UK.
Backpropagation (the process of a network learning from its mistakes) made it possible to train multi-layer neural nets, laying the groundwork for modern machine learning.
A BRIEF HISTORY OF AI
2017
2015-16
2012
1997
Deep Learning
Computing Muscle
Link
Link
Link
Link
The AlexNet neural network demonstrated unparalleled performance correctly classifying images on ImageNet, the largest image dataset at the time. This milestone is considered the birth of deep learning, when machines learn by mimicking human brain structure to analyze data.
The Transformer neural network architecture made language models faster to train and much better at understanding context, setting the stage for today’s AI.
IBM's Deep Blue computer beat world chess champion Garry Kasparov, an iconic moment showing what high-speed computations and specialized hardware could do.
Deep reinforcement learning matched or exceeded human play (DQN on Atari and AlphaGo defeating Go champion Lee Sedol) by learning to make better decisions through a reward/penalty system.
A BRIEF HISTORY OF AI
TOMORROW
TODAY
2023-25
2020-22
Multimodality
Generative AI
Link
Link
Link
Link
Rapid progress in foundation models (AI models trained on vast datasets using self-supervised learning) and AI agents that use logic, planning, and take action using external tools to achieve goals without constant human supervision.
Research now balances capability with safety, reliability, and responsible use - areas that are advancing alongside the models themselves. The field of AI is dynamic and constantly evolving.
Diffusion models (generative AI that can create images from scratch) enabled high-quality image generation. Chat-style language models went mainstream with ChatGPT.
State-of-the-art systems are multimodal (text, images, audio, and more), can generate content, write code, reason, and increasingly use tools. AI trends are tracked annually by the AI Index Report.
History of AI Timeline
jamieweaver
Created on February 26, 2026
Start designing with a free template
Discover more than 1500 professional designs like these:
View
Timeline Lines Mobile
View
Major Religions Timeline
View
Timeline Flipcard
View
Timeline video
View
Images Timeline Mobile
View
Sport Vibrant Timeline
View
Decades Infographic
Explore all templates
Transcript
A BRIEF HISTORY OF AI
1986
1980
1973
1958
1956
Early Criticism
The AI Comeback
Foundations
Link
Link
Link
Link
Link
Frank Rosenblatt's perceptron demonstrated a machine could learn simple patterns, inspiring early neural network research.
Expert systems capable of solving complex problems (like DEC's XCON rule-based program) delivered practical value in business, renewing interest in AI.
Researchers began asking if machines could "think." A group of scientists gathered at Dartmouth to research "machine learning" and coined the term "artificial intelligence."
The controversial Lighthill Report criticized AI research, stating that while AI could solve simple problems, it was not scalable to solve more complex real-world tasks. This led to a dramatic decrease in AI research in the UK.
Backpropagation (the process of a network learning from its mistakes) made it possible to train multi-layer neural nets, laying the groundwork for modern machine learning.
A BRIEF HISTORY OF AI
2017
2015-16
2012
1997
Deep Learning
Computing Muscle
Link
Link
Link
Link
The AlexNet neural network demonstrated unparalleled performance correctly classifying images on ImageNet, the largest image dataset at the time. This milestone is considered the birth of deep learning, when machines learn by mimicking human brain structure to analyze data.
The Transformer neural network architecture made language models faster to train and much better at understanding context, setting the stage for today’s AI.
IBM's Deep Blue computer beat world chess champion Garry Kasparov, an iconic moment showing what high-speed computations and specialized hardware could do.
Deep reinforcement learning matched or exceeded human play (DQN on Atari and AlphaGo defeating Go champion Lee Sedol) by learning to make better decisions through a reward/penalty system.
A BRIEF HISTORY OF AI
TOMORROW
TODAY
2023-25
2020-22
Multimodality
Generative AI
Link
Link
Link
Link
Rapid progress in foundation models (AI models trained on vast datasets using self-supervised learning) and AI agents that use logic, planning, and take action using external tools to achieve goals without constant human supervision.
Research now balances capability with safety, reliability, and responsible use - areas that are advancing alongside the models themselves. The field of AI is dynamic and constantly evolving.
Diffusion models (generative AI that can create images from scratch) enabled high-quality image generation. Chat-style language models went mainstream with ChatGPT.
State-of-the-art systems are multimodal (text, images, audio, and more), can generate content, write code, reason, and increasingly use tools. AI trends are tracked annually by the AI Index Report.