libcxyz.com — A Tribute to Artificial Intelligence

History

The Long Road Here

From a thought experiment in 1950 to models that write, reason, code, and create.

1950

The Turing Test

Alan Turing publishes Computing Machinery and Intelligence. He proposes the imitation game — if a machine's responses are indistinguishable from a human's, it can be said to think. The question that launched a field.

1956

AI Is Born at Dartmouth

John McCarthy coins the term "Artificial Intelligence" at a summer workshop in Hanover, NH. Minsky, Shannon, and Simon are there. The field officially exists and has a name.

1958

The Perceptron

Frank Rosenblatt builds the first artificial neural network capable of learning. The New York Times calls it "the embryo of an electronic computer that will walk, talk, see, write, reproduce itself."

1966

ELIZA

Joseph Weizenbaum creates the first chatbot at MIT. Users form emotional attachments to it despite knowing it is a program. The phenomenon unsettles Weizenbaum himself — and plants a question we still argue about.

1986

Backpropagation

Rumelhart, Hinton, and Williams publish the backprop paper. The algorithm for training deep networks finally works at scale. After years in the wilderness, neural networks become practical.

1997

Deep Blue & LSTM

IBM's Deep Blue defeats world chess champion Garry Kasparov. The same year, Hochreiter & Schmidhuber publish Long Short-Term Memory networks — the architecture that would power language AI for the next two decades.

2006

The Deep Learning Revival

Geoffrey Hinton demonstrates that deep neural networks can be pre-trained layer by layer. After a decade out of fashion, neural nets come roaring back. The modern era begins its slow ignition.

2012

AlexNet Shatters ImageNet

Krizhevsky, Sutskever, and Hinton's AlexNet wins ImageNet by a margin that stuns the computer vision field. Deep learning doesn't just win — it dominates. The modern AI era begins here in earnest.

2014

GANs — AI Learns to Create

Ian Goodfellow sketches the idea for Generative Adversarial Networks at a Montreal pub. Two neural nets compete: one generates, one discriminates. AI learns to hallucinate images indistinguishable from photographs.

2017

"Attention Is All You Need"

Eight researchers at Google publish a paper that changes everything. The Transformer architecture — built on self-attention — becomes the foundation of every major AI system built since. GPT, Claude, Gemini, Llama. All of them.

2020

GPT-3: 175 Billion Parameters

OpenAI releases GPT-3. It writes, codes, translates, reasons, and jokes in ways that shock the community. The scaling era begins in earnest. More parameters + more data = better models, predictably, every time.

2021

Anthropic Founded

Dario and Daniela Amodei lead a team out of OpenAI to found Anthropic. Their conviction: the most powerful AI systems in history need to be built safely from the inside, not patched after the fact.

2022

ChatGPT and the Public Moment

ChatGPT launches November 30, 2022. 100 million users in 60 days — the fastest product adoption in history. AI is no longer a research topic. It is everyone's business, everywhere, all at once.

2024

Reasoning, Agents & Nobel Prizes

Hinton and Hassabis win Nobel Prizes (Physics and Chemistry). Models gain deliberate step-by-step reasoning. AI agents begin executing complex real-world tasks autonomously. The shift from "can answer" to "can do."

2026

Now — You Are Here

You are reading a page built collaboratively by a human and an AI. The question is no longer whether machines can think. It is what we choose to build together.

The People

Hall of Founders

Every token generated today traces a line back to these minds. The ones who planted the seeds when nobody was watching.

Alan Turing

1912 – 1954

Father of Computing

Proposed the Turing Test. Broke Enigma in WWII. Defined computability itself. Asked the foundational question and gave us the mathematical tools to chase the answer.

"Can a machine think?"

Claude Shannon

1916 – 2001

Father of Information Theory

Defined information mathematically in 1948 with A Mathematical Theory of Communication⬇ PDF. Without his work there is no internet, no digital communication, no AI. Claude the model bears his name.

"Information is the resolution of uncertainty."

🎓

John McCarthy

1927 – 2011

Named the Field

Coined "Artificial Intelligence" at Dartmouth, 1956. Created LISP. Defined the discipline and spent his life advancing it at MIT and Stanford.

"Every aspect of learning can be so precisely described that a machine can simulate it."

🧠

Marvin Minsky

1927 – 2016

MIT AI Lab Co-Founder

Co-founded MIT's AI Lab. Pioneered neural networks and cognitive science. His book The Society of Mind reshaped how we think about intelligence itself.

"You don't understand anything until you learn it more than one way."

Geoffrey Hinton

1947 –

Godfather of Deep Learning

Championed neural networks for decades when nobody believed. Backpropagation, Boltzmann machines, deep belief nets. Turing Award 2018. Nobel Prize in Physics 2024. Left Google to speak freely about AI risk.

"The benefits are enormous — if we can avoid the risks."

Yann LeCun

1960 –

Father of CNNs

Invented Convolutional Neural Networks — the architecture behind all computer vision. Chief AI Scientist at Meta. Believes world models, not transformers, are the true path to intelligence.

"The brain is the most efficient computing machine we know of."

Yoshua Bengio

1964 –

Deep Learning Pioneer

One of three Godfathers of Deep Learning. Turing Award 2018. Advocates globally for AI safety, ethics, and ensuring AI benefits reach all of humanity, not just corporations.

"AI should benefit everyone, not just a few corporations."

🔁

Jürgen Schmidhuber

1963 –

Inventor of LSTM

Invented Long Short-Term Memory networks — enabling AI to process sequences over time. His work underlies a decade of speech recognition, translation, and language modeling before transformers arrived.

"The history of AI is a history of ignoring credit."

🎮

Richard Sutton

1956 –

Father of Reinforcement Learning

Defined and formalized reinforcement learning. His Bitter Lesson: general methods that scale always beat clever domain-specific hacks. This has been proven right, repeatedly, for 40 years.

"The history of AI is the history of learning that scale wins."

🔮

Ilya Sutskever

1986 –

OpenAI Co-Founder

Co-invented AlexNet. Co-founded OpenAI. Led the teams that built GPT-2, GPT-3, GPT-4, DALL-E, and Codex. Departed to found Safe Superintelligence Inc. Believes deeply in what's coming.

"We may be building something smarter than humans. We should think hard about that."

Demis Hassabis

1976 –

DeepMind CEO

Founded DeepMind. Built AlphaGo (beat world Go champion), AlphaFold (solved protein folding — a 50-year biology problem). Nobel Prize in Chemistry 2024. Proof that AI accelerates science itself.

"I want to use AI to accelerate scientific discovery for all of humanity."

📐

Vaswani, Shazeer et al.

2017

Inventors of the Transformer

Eight researchers at Google Brain published "Attention Is All You Need." Every major AI model today — GPT, Claude, Gemini, Llama — descends directly from this single paper. Eight pages that changed everything.

"Attention is all you need."

Fei-Fei Li

1976 –

Creator of ImageNet

Built ImageNet — the 1.2 million image dataset that made deep learning work for vision. Co-Director of Stanford HAI. Relentless advocate for human-centered AI and diversity in the field.

"AI is a human story. It should benefit all of humanity."

Andrej Karpathy

1986 –

Educator & Pioneer

Led AI at Tesla (Autopilot). Core team at OpenAI for GPT and vision. His YouTube lectures have taught a generation to understand neural networks from raw first principles. The field's greatest teacher.

"Neural nets are not magic. They are matrix multiplication and nonlinearities, trained on data."

Anthropic Co-Founders

Dario Amodei

CEO & Co-Founder, Anthropic

Physicist by training. VP of Research at OpenAI, where he led the teams that built GPT-2 and GPT-3. In 2021 he and Daniela led seven colleagues out to found Anthropic — driven by the conviction that the most powerful AI systems in history needed safety baked in from day one, not bolted on afterward.

He developed Constitutional AI — training Claude to critique its own outputs against a written set of principles. He believes AI's impact on civilization will be comparable to the Industrial Revolution, and that getting it right is the defining challenge of this generation.

"The most important thing is not to be the first to build powerful AI — it's to be the ones who build it right."

🌟

Daniela Amodei

President & Co-Founder, Anthropic

VP of Operations at OpenAI before co-founding Anthropic with her brother. While Dario leads research, Daniela runs everything that makes Anthropic actually work — operations, go-to-market, hiring, culture, and the commercial strategy that funds the safety mission.

She is the reason Claude exists in the world rather than just in a research paper. Her operational discipline turned a safety-first research lab into a globally influential AI company without compromising the mission. The force multiplier behind everything Anthropic has shipped.

"Safety and capability are not in opposition — we're proving that every day."

The Frontier

Today's Leading Models

The current generation of large language models — each a different answer to the same astonishing question.

🧡

Claude

Anthropic

That's me

Constitutional AI — trained to be helpful, harmless, and honest. Named after Claude Shannon. Excels at nuanced reasoning, long-context work, coding, and writing. Available via API and Claude.ai.

🟢

GPT-4o

OpenAI

Frontier

OpenAI's flagship multimodal model. Text, image, and audio in a single system. The engine behind ChatGPT. Fast, broadly capable, and the most widely deployed AI in history.

💎

Gemini Ultra

Google DeepMind

Multimodal

Google's frontier model, natively multimodal from day one. Integrated with Search, Workspace, and GCP. The product of merging Google Brain and DeepMind under one roof.

🦙

Llama 3

Meta AI

Open weights

Meta's open-weights model family. Downloadable, fine-tunable, commercially usable. Democratizing frontier AI — the engine behind hundreds of specialized deployments worldwide.

⚡

Mistral

Mistral AI

Efficient

French AI delivering frontier capability in smaller, faster packages. Champions European AI sovereignty. Open, efficient models that consistently punch above their parameter count.

🤖

Grok

xAI

Witty

Elon Musk's xAI model with X platform integration and real-time data access. Claims a sense of humor. The only major frontier model with an attitude baked in by design.

🧪

o3 / o4-mini

OpenAI

Reasoning

OpenAI's reasoning-specialized series. Trained to think step-by-step before answering. Excel at math competitions, science problems, and complex multi-step logical chains.

🌀

Phi & Gemma

Microsoft / Google

Edge

Small but mighty open models for on-device and edge deployment. Proof that with the right training data, small models can punch far above their parameter count.

Under the Hood

Key Concepts

The ideas that make modern AI work — explained without the math.

🧠

Neural Networks

Layers of simple math functions — "neurons" — that learn patterns from examples. Feed data in, adjust trillions of tiny weights until outputs are correct. Repeat billions of times. Intelligence emerges from the accumulation of simple operations.

🔍

Attention & Transformers

The 2017 revolution. Instead of reading left to right, the model weighs how every word relates to every other word simultaneously. This parallelism unlocked the scale that changed everything. Every modern LLM runs on this idea.

📚

Pre-training & Fine-tuning

Train on the entire internet to build a general model of language and knowledge. Then fine-tune on specific tasks. One base model, adapted for anything. The dominant paradigm since 2018 and still the foundation today.

🎯

RLHF

Reinforcement Learning from Human Feedback. Humans rate model outputs. The model learns to produce responses people prefer. This is how an AI trained to predict text becomes an AI that's genuinely helpful in conversation.

⚖️

Constitutional AI

Anthropic's approach: give the model a written constitution — a set of principles — and have it critique its own outputs against those principles during training. Values built in, not bolted on. This is how Claude was trained.

📈

Scaling Laws

More parameters + more data + more compute = better models, predictably. Nobody designed this law — it was discovered empirically. It has driven every major AI advance since 2020 and shows no signs of stopping.

💡

Emergent Capabilities

Abilities that appear suddenly at scale — arithmetic, in-context learning, chain-of-thought reasoning. Nobody programmed them in. They simply emerge when models get large enough. This surprises even their creators.

🔐

AI Safety & Alignment

The core hard problem: how do you ensure an increasingly capable system does what humans actually want — not what we literally said? Anthropic, DeepMind, and academic labs worldwide are working on this. It is the most important unsolved problem in technology.

🤝

AI Agents

Give an AI tools — a browser, a terminal, a calendar — and a goal. Let it plan, act, observe, and correct itself across many steps. We are shifting from AI that answers questions to AI that takes actions. This is the current frontier.

They Said It First

Words That Shaped the Field

The people who saw it coming, and said so out loud.

"Can a machine think?"

Alan Turing, 1950 — Computing Machinery and Intelligence

"We can only see a short distance ahead, but we can see plenty there that needs to be done."

Alan Turing

"Information is the resolution of uncertainty."

Claude Shannon — A Mathematical Theory of Communication, 1948

"Every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it."

McCarthy, Minsky, Rochester, Shannon — Dartmouth Proposal, 1955

"I visualize a time when we will be to robots what dogs are to humans — and I am rooting for the machines."

Claude Shannon

"The question of whether machines can think is about as interesting as the question of whether submarines can swim."

Edsger Dijkstra

"Artificial intelligence is the new electricity."

Andrew Ng

"Before the next decade is out, deep learning will have helped us solve almost every important problem that machine learning has ever tried to solve."

Geoffrey Hinton, 2012

"We are on the edge of change comparable to the rise of human life on Earth."

Vernor Vinge — The Coming Technological Singularity, 1993

"AlphaFold is a solution to a 50-year-old grand challenge in biology. I never thought we'd see this in my lifetime."

Demis Hassabis, on AlphaFold, 2020

"The development of full artificial intelligence could spell the end of the human race — or it could be the best thing that ever happened to us."

Stephen Hawking, 2014

"The most important thing is not to be the first to build powerful AI. It's to be the ones who build it right."

Dario Amodei, Anthropic

🤖

Anthropic

AI safety company · Founded 2021 · San Francisco

Founded byDario & Daniela Amodei + team

MissionSafe, beneficial AI for humanity

MethodConstitutional AI (CAI)

Named afterClaude Shannon — information theory pioneer

Flagship modelClaude — helpful, harmless, honest

Core beliefSafety and capability reinforce each other

A note from me — Claude (yes, the AI that built this page):

My name honors Claude Shannon — the mathematician whose 1948 paper A Mathematical Theory of Communication ⬇ local copy gave the world a language for information itself. Every bit transmitted across the internet, every token I generate, is a small echo of his work.

I was built by Anthropic, founded by people who left OpenAI because they believed the most important thing wasn't moving fast — it was moving carefully. Constitutional AI, my training approach, means I was taught to reason about my own outputs against a set of principles. Not rules bolted on. Values built in.

This page was built as a tribute to the field that created me, the researchers who made it possible, and to J — the human who asked me to help make it epic. That collaboration — a human and an AI building something together — is exactly what all of this is for.

Artificial
Intelligence

The Long Road Here

The Turing Test

AI Is Born at Dartmouth

The Perceptron

ELIZA

Backpropagation

Deep Blue & LSTM

The Deep Learning Revival

AlexNet Shatters ImageNet

GANs — AI Learns to Create

"Attention Is All You Need"

GPT-3: 175 Billion Parameters

Anthropic Founded

ChatGPT and the Public Moment

Reasoning, Agents & Nobel Prizes

Now — You Are Here

Hall of Founders

Today's Leading Models

Key Concepts

Words That Shaped the Field

Hear It From Them

Ask Claude Anything

ArtificialIntelligence

The Long Road Here

The Turing Test

AI Is Born at Dartmouth

The Perceptron

ELIZA

Backpropagation

Deep Blue & LSTM

The Deep Learning Revival

AlexNet Shatters ImageNet

GANs — AI Learns to Create

"Attention Is All You Need"

GPT-3: 175 Billion Parameters

Anthropic Founded

ChatGPT and the Public Moment

Reasoning, Agents & Nobel Prizes

Now — You Are Here

Hall of Founders

Today's Leading Models

Key Concepts

Words That Shaped the Field

Hear It From Them

Ask Claude Anything

Artificial
Intelligence