🧠 51 researchers
A chronological list of the key AI researchers, from the 1950s theorists who named the field to the labs shipping today’s large language models. Each is placed by the landmark that mattered—the Turing Test, the perceptron, AlexNet, the Transformer—with a one-line verdict and a link to the full profile. No single person invented AI; this is the relay, in order.
They named the field, wrote its theory, and built the first thinking machines—decades before the hardware could keep up.
A science-fiction writer, not an engineer—but his Three Laws of Robotics framed how everyone since has argued about controlling machines.
Founded cybernetics—feedback and control as one theory for animals and machines—and warned early that automation could go badly for the people it replaced.
Information theory gave every later system its unit—the bit. Also wrote the first serious paper on programming a computer to play chess.
Defined computation itself, then asked “Can machines think?” and proposed the test that still bears his name. The whole field starts here.
Coined the term “artificial intelligence” for the 1956 Dartmouth workshop that launched the field, and invented Lisp, its first native language.
Built the first program that could prove mathematical theorems—arguably the first true AI—and made the case that thinking is symbol manipulation.
Built the perceptron, the first machine that learned from examples—the direct ancestor of every neural network. Minsky’s critique then froze the idea for a decade.
Coined “machine learning” and wrote a checkers program that improved by playing itself—self-play, sixty years before AlphaZero.
Co-founded the MIT AI Lab and shaped symbolic AI for a generation—then his book on the perceptron’s limits helped trigger the first AI winter.
Through two “AI winters” they kept betting on neural networks while the field looked away. Backpropagation, CNNs, LSTMs—the machinery of the coming revolution.
Described backpropagation—the algorithm that trains every deep network—in his PhD thesis, a decade before the field noticed.
Co-invented the Boltzmann machine with Hinton and built the bridge between neuroscience and deep learning that both fields still cross.
Argued intelligence needs a body, not a symbol database—“the world is its own best model.” Later put a Roomba in millions of homes.
The 1986 paper he co-wrote made backpropagation practical and relaunched neural networks. The psychologist who gave connectionism its rigor.
The “godfather of deep learning.” Kept neural nets alive through both winters; his lab’s 2012 AlexNet won the field. Quit Google in 2023 to warn about the risk. Nobel 2024.
Gave AI a rigorous way to reason under uncertainty (Bayesian networks), then spent his later career arguing today’s AI still can’t handle cause and effect.
Invented the convolutional network that read bank checks in the 1990s and now underlies computer vision. Meta’s chief AI scientist, and the field’s loudest skeptic of LLM hype.
Co-invented the LSTM that made sequence models work and powered a decade of translation and speech. Also famous for insisting he invented nearly everything first.
Wrote the book on reinforcement learning—the training method behind AlphaGo and RLHF. His essay “The Bitter Lesson” became the manifesto for scaling.
Laid the groundwork for neural language models and attention; shared the 2018 Turing Award with Hinton and LeCun. Now among the most vocal on catastrophic risk.
Big data plus GPUs plus deep nets. ImageNet and AlexNet flipped the field in a single year; DeepMind beat the game of Go.
Built ImageNet, the giant labeled dataset whose annual contest set off the deep-learning boom. Proof that data, not just algorithms, moved the field.
Co-founded DeepMind and popularized the term “artificial general intelligence.” Argued for taking AGI timelines seriously years before it was fashionable.
Co-founded Google Brain and, through Coursera, taught machine learning to millions. Did more than anyone to turn a niche into a profession.
Co-built AlexNet, the system that converted the whole field to deep learning overnight. Co-founded OpenAI, championed scaling, then left in 2024 to build “safe superintelligence.”
Invented generative adversarial networks—two nets competing to fake and detect—reportedly sketched in a bar. The idea behind a decade of synthetic images.
Chess prodigy turned neuroscientist turned founder of DeepMind. AlphaGo beat Go in 2016; AlphaFold cracked protein folding and won him a Nobel in 2024.
Led AlphaGo and then AlphaZero, which learned Go, chess, and shogi from nothing but the rules—the cleanest proof of Sutton’s bitter lesson.
The Transformer, the scaling bet, and the labs—OpenAI, Anthropic, DeepMind, DeepSeek—that turned research into products used by hundreds of millions.
Google’s legendary systems engineer, co-founder of Google Brain, and the reason the infrastructure existed to train models at scale. Now chief scientist.
Co-founded OpenAI and, as CEO, shipped ChatGPT—the fastest product to 100 million users ever. Fired and reinstated in five days in 2023; the face of the boom.
OpenAI’s president and founding engineer—the one who actually built and shipped the infrastructure behind GPT. Left with Altman in 2023, came back.
Co-founded and funded OpenAI, left after a fight for control, then launched xAI to compete with it—and sued it. AI’s loudest and most litigious backer.
Founding OpenAI member, led Tesla’s self-driving vision, and the field’s best teacher—his lectures taught a generation how neural nets actually work.
Co-founded DeepMind, then Inflection, and now runs Microsoft AI. The co-founder who focused on product and policy rather than the math.
Founder-CEO of Nvidia, whose GPUs turned out to be the hardware every neural network runs on. Bet the company on AI compute and became its landlord.
Eight Google researchers whose 2017 paper introduced the Transformer—the architecture under every large language model since. The single most important paper of the era.
OpenAI co-founder whose reinforcement-learning algorithms (PPO, RLHF) are how ChatGPT was tuned to follow instructions. The method behind the manners.
Lead author of BERT, the model that made Transformers dominate language understanding and quietly went into Google Search for billions of queries.
Lead author of the original GPT papers. The quiet researcher whose work turned the Transformer into the generative models everyone now uses.
Left OpenAI over safety differences to co-found Anthropic and build Claude. Argues you can be at the frontier and take the risks seriously at once.
As OpenAI’s CTO she shipped ChatGPT, DALL·E, and GPT-4, then left in 2024 to found her own lab. The operator who turned research into the product.
Quant-fund founder who built DeepSeek, whose open models matched Western frontier labs at a fraction of the cost and rattled the market in early 2025.
Founder of Baidu, which shipped China’s first major ChatGPT rival, Ernie Bot. Bet China’s search giant on the model race early.
Ran Google and Microsoft in China, wrote “AI Superpowers,” then founded 01.AI. The clearest translator between the American and Chinese AI worlds.
The people asking whether we can control what we’re building—and the ones arguing the hype outruns the reality.
Founded the field of AI alignment before it had a name, arguing for decades that misaligned superintelligence is an extinction risk. The original doomer.
His book “Superintelligence” put AI existential risk on the agenda of researchers and CEOs alike—the paperclip thought experiment made the danger concrete.
Pioneered learning from human feedback as an alignment tool, then left to run US government AI safety testing. The bridge between the doomers and the labs.
The field’s most persistent critic: argues deep learning alone can’t reach real intelligence and that the hype is dangerous. Often dismissed, occasionally vindicated.
Her “Gender Shades” study proved commercial face recognition failed on darker-skinned women—hard evidence that turned algorithmic bias from theory into law.
Co-wrote the standard AI textbook, then argued the whole field is built on a broken goal—machines pursuing fixed objectives—and needs redesigning around human uncertainty.
Co-wrote the “stochastic parrots” paper warning of bias and cost in large language models—and was pushed out of Google over it. The face of AI ethics.
Co-founded EleutherAI to reproduce GPT in the open, then turned sharply toward safety, arguing the open models he helped start are part of the danger.
Co-led OpenAI’s superalignment team, then quit publicly in 2024 saying safety had “taken a back seat to shiny products,” and joined Anthropic.
No single person. John McCarthy coined the term “artificial intelligence” in 1956 and organized the Dartmouth workshop that founded the field; Alan Turing laid its theoretical basis in 1950; and Marvin Minsky, Allen Newell, and Herbert Simon built its first working programs. Modern AI—deep learning—traces instead to Geoffrey Hinton, Yann LeCun, and Yoshua Bengio.
The title usually goes to John McCarthy, who named the field, or shares it with Marvin Minsky. For modern deep learning, Geoffrey Hinton is called the “godfather of AI.” Alan Turing is considered the father of computing and theoretical AI.
OpenAI (2015) was co-founded by Sam Altman, Greg Brockman, Ilya Sutskever, Elon Musk, and John Schulman, among others. DeepMind (2010) by Demis Hassabis, Shane Legg, and Mustafa Suleyman. Anthropic (2021) by Dario Amodei and a group who left OpenAI over safety.
This roster covers the field’s most consequential figures across five eras—from the 1940s founders to today’s LLM builders and the safety researchers questioning them. It is a curated list of the people who moved the field, not a complete census.