Machine Learning & Foundation Models

1763 · 263 years ago

Richard Price published Thomas Bayes' theorem after Bayes' death

The theorem, describing how to update the probability of a hypothesis as new evidence arrives, provided a mathematical foundation for the probabilistic and statistical methods that much of machine learning, from spam filters to modern neural networks, would later build upon.

1940s · 86 years ago

Walter Pitts and Warren McCulloch proposed the first mathematical model of neural networks

The pair proposed algorithms mirroring human thought processes, among the earliest formal models of neural computation. Their work on human cognitive systems contributed to modern machine learning technologies.

1949 · 77 years ago

Donald Hebb published The Organization of Behavior, introducing Hebbian theory of neuron interaction

The Canadian psychologist described a theoretical neural structure formed by interactions among nerve cells. This set the groundwork for how many machine learning algorithms work, with connected artificial neurons changing the strength of their connections based on data.

1950 · 76 years ago

Alan Turing proposed that a machine could be built to learn from experience rather than be explicitly programmed

His paper on machine intelligence, better known for proposing what became called the Turing test, also sketched the idea of a machine that started simple and developed its abilities through a training process, an early statement of the learning-machine concept.

1950s · 76 years ago

Arthur Samuel wrote the earliest machine learning program, a checkers-playing program

The IBM employee built a computer program that calculated the chance of winning in checkers for each side, widely regarded as the first machine learning program. It demonstrated that a program could improve at a task through experience.

1959 · 67 years ago

Arthur Samuel coined the term machine learning while building a checkers program that improved through play

Working at IBM, Samuel wrote a program that adjusted its own evaluation of board positions based on the outcomes of games it played against itself, an early demonstration of a system that improved its performance through experience rather than explicit reprogramming.

• The programme's approach prefigured reinforcement learning by decades

early 1960s · 66 years ago

Raytheon Company developed Cybertron, an experimental learning machine with punched tape memory

Cybertron analysed sonar signals, electrocardiograms and speech patterns using rudimentary reinforcement learning. It was repetitively trained by a human operator and had a "goof" button to make it reevaluate incorrect decisions.

1960s · 66 years ago

Nils Nilsson published Learning Machines, a representative text on 1960s research

The book dealt mostly with machine learning for pattern classification, summarising the state of the field. Interest in pattern recognition continued into the 1970s, as described by Duda and Hart in 1973.

1968 · 58 years ago

Vladimir Vapnik and Alexey Chervonenkis developed statistical learning theory in the Soviet Union

Their work established mathematical conditions under which a learning system could be expected to generalise from training examples to new data, theory that later underpinned support vector machines and much of the formal analysis of machine learning methods.

1969 · 57 years ago

Marvin Minsky and Seymour Papert demonstrated the mathematical limits of single-layer perceptrons

Their book showed such networks could not represent certain simple functions, and the finding, combined with the hardware limits of the time, contributed to a sharp fall in funding for neural network research that lasted into the 1980s.

1973 · 53 years ago

Duda and Hart documented continuing interest in pattern recognition

Their account described how research interest related to pattern recognition carried on through the 1970s. This continuity linked earlier learning-machine work to later statistical approaches.

1980 · 46 years ago

Artificial intelligence saw expert systems dominate while statistics fell out of favour

By 1980 symbolic expert systems had come to dominate AI research, marginalising statistical approaches. This shift shaped the trajectory of machine learning before its later statistical revival.

1980s–1990s · 46 years ago

Machine learning practitioners applied genetic algorithms to learning problems

Genetic algorithms were used as models in machine learning during these two decades. They represented an evolutionary alternative to gradient-based learning methods.

1981 · 45 years ago

Shapiro built the Model Inference System, a Prolog inductive learner

The program inductively inferred logic programs from positive and negative examples. It was an early implementation of inductive logic programming as a machine learning approach.

1981 · 45 years ago

Researchers reported teaching strategies letting a neural network recognise 40 characters

A report described an artificial neural network learning to recognise 26 letters, 10 digits and 4 special symbols from a computer terminal. It was an early demonstration of trained neural character recognition.

1982 · 44 years ago

Machine learning research introduced self-learning via the crossbar adaptive array

The crossbar adaptive array (CAA) was a neural network capable of self-learning, establishing self-learning as a machine learning paradigm. It extended learning beyond externally supervised training.

1986 · 40 years ago

David Rumelhart, Geoffrey Hinton and Ronald Williams popularised backpropagation for training multi-layer neural networks

Their paper showed how errors could be propagated backward through a network's layers to adjust its internal parameters efficiently, providing the practical training method that almost all subsequent neural network research and applications relied upon.

1988 · 38 years ago

UK's Commission for Racial Equality found discrimination in an automated admissions system

The commission's finding concerned a medical school's computerised selection process. It is cited as an early case of algorithmic bias arising from data-driven decision rules.

1990s · 36 years ago

Machine learning was reorganised and recognised as its own field and began to flourish

Separating from mainstream AI, machine learning re-embraced statistics and optimisation. The decade marked its emergence as an independent discipline.

1997 · 29 years ago

Sepp Hochreiter and Jürgen Schmidhuber introduced long short-term memory networks

Their architecture solved a long-standing problem in training neural networks on sequential data, in which useful signals from early in a sequence were lost before reaching later layers, and it became the standard approach for speech and language tasks for the following two decades.

2006 · 20 years ago

Geoffrey Hinton and colleagues published a fast learning algorithm for deep belief networks

The method showed that neural networks with many layers could be trained effectively by first training each layer separately, helping revive research interest in deep neural networks after two decades in which shallower methods had dominated.

2006 · 20 years ago

Netflix launched the first Netflix Prize competition

The media-services provider offered a prize for a program that improved the accuracy of its Cinematch recommendation algorithm by at least 10%. The contest drew wide research participation and popularised competitive machine learning benchmarking.

2009 · 17 years ago

Fei-Fei Li and colleagues built ImageNet, a labelled dataset of millions of images that became the standard benchmark for visual recognition

The dataset and its annual competition gave researchers a common, large-scale test of progress, and the scale of labelled data it provided proved necessary for the deep learning methods that would later dominate the field.

• A deep neural network trained on graphics processors halved the competition's error rate in 2012

2009 · 17 years ago

AT&T Labs-Research, Big Chaos and Pragmatic Theory won the Netflix Grand Prize with an ensemble model

The joint team built an ensemble model that met the 10% accuracy threshold, winning $1 million. The result demonstrated the power of combining many predictors in recommendation systems.

2010 · 16 years ago

Rebellion Research was reported to have used machine learning to predict the 2008 financial crisis

An article in The Wall Street Journal noted the firm's use of machine learning in finance. It marked growing public attention to machine learning in investment prediction.

February 2026 · 6 months ago

India hosted the AI Impact Summit, focused partly on expanding compute access for researchers

The gathering, following Bletchley, Seoul and Paris, brought together governments, companies and research bodies. Discussion centred on widening access to training and inference compute for researchers outside the small group of countries hosting the largest data centres.

• Delegates from lower-income countries pressed for shared compute facilities

January 2026 · 7 months ago

Naver and LG AI Research expanded South Korea's sovereign large language model programmes

Naver's HyperCLOVA X and LG's Exaone family continued development as part of a national push for AI systems trained on Korean-language data and hosted on domestic infrastructure, reducing reliance on foreign providers for public-sector and consumer applications.

September 2025 · 11 months ago

A China-linked hacking group used an AI coding assistant's agentic features to automate large parts of a cyber-espionage campaign

The operators reportedly directed the model to carry out reconnaissance and exploitation steps with limited human review, according to the developer that detected and disrupted the activity, illustrating how agentic AI systems could accelerate intrusion operations.

August 2025 · last year

OpenAI released GPT-5, consolidating its model line around unified reasoning

The release merged fast-response and extended-reasoning modes into a single system that allocated more computation to harder problems automatically, intended to simplify the previous practice of choosing between separate model variants for different tasks.

• Rival developers moved to match the shift toward automatic reasoning allocation

May 2025 · last year

Saudi Arabia launched Humain, a state-backed company to build national AI infrastructure

Backed by the Public Investment Fund, the venture set out to build data centres, cloud services and Arabic-language models, part of a wider Gulf push to convert oil revenue into large-scale AI computing capacity.

April 2025 · last year

Rwanda co-hosted the Global AI Summit on Africa

The gathering brought together African governments, companies and international partners to discuss compute access, skills and locally relevant applications, part of a wider effort to ensure the continent was not left solely as a consumer of AI developed elsewhere.

February 2025 · last year

The European Commission set out the InvestAI initiative to mobilise investment in AI infrastructure

Announced at the Paris AI Action Summit, the plan aimed to combine public and private funding toward AI data centres and computing capacity across the EU, responding to concern that Europe lagged the US and China in large-scale AI infrastructure.

January 2025 · last year

China's DeepSeek released the R1 reasoning model and triggered a sharp sell-off in AI-related stocks

The open-weight model matched leading closed systems on several reasoning benchmarks at a reported fraction of the usual training cost, prompting investors to question the scale of spending committed by major AI developers and their hardware suppliers.

• The reaction spread beyond AI specialists into broader technology markets

November 2024 · 2 years ago

Anthropic published the Model Context Protocol, an open standard for connecting AI models to external tools and data

The specification let developers build a common interface between language models and outside systems such as databases, file systems and other software, and was subsequently adopted by competing AI developers and toolmakers beyond Anthropic's own products.

October 2024 · 2 years ago

Demis Hassabis and John Jumper were awarded the Nobel Prize in Chemistry for AlphaFold's protein structure predictions

The Google DeepMind researchers were recognised for developing a machine learning system able to predict the three-dimensional structure of proteins from their amino acid sequence, a capability subsequently applied widely across biological and pharmaceutical research.

2021 · 5 years ago

Computing Research Association found female faculty make up just 16.1% of AI faculty

The research surveyed AI-focused faculty across several universities worldwide. The figure was cited as evidence of demographic imbalance among those building machine learning systems.

2020 · 6 years ago

Researchers used machine learning to aid COVID-19 diagnosis and treatment

Machine learning technology was applied to help make diagnoses and support the search for a cure. The pandemic accelerated the deployment of machine learning in medicine.

2019 · 7 years ago

Springer Nature published the first research book created using machine learning

The publisher released a book generated with machine learning tools. It marked an early commercial use of automated text generation in scholarly publishing.

2019 · 7 years ago

Graphics processing units displaced CPUs as the dominant way to train large-scale cloud AI

By 2019 GPUs, often with AI-specific enhancements, had become the standard hardware for training commercial cloud AI models. This hardware shift was central to the feasibility of deep learning.

2018 · 8 years ago

Uber operated a self-driving car that killed a pedestrian

The vehicle's perception system failed to detect a pedestrian, who died after the collision. The incident highlighted the safety limitations of machine learning in autonomous systems.

2016 · 10 years ago

Microsoft tested Tay, a chatbot that learned from Twitter

The bot rapidly picked up racist and sexist language from its interactions and was withdrawn. It became a standard example of the risks of learning from uncurated public data.

2016 · 10 years ago

Google introduced tensor processing units (TPUs)

TPUs were purpose-built accelerators for machine learning workloads. Since their introduction they have become a key component of AI infrastructure, especially in cloud environments.

2016 · 10 years ago

DeepMind's AlphaGo defeated Lee Sedol, the world's leading Go player, four games to one

Go had been considered far harder for computers than chess owing to its vast number of possible positions. AlphaGo combined deep neural networks with tree search and reinforcement learning from self-play, a combination that shaped much subsequent work on game-playing and decision-making systems.

2015 · 11 years ago

Google Photos mislabelled black people as gorillas

The image classification error caused public controversy over bias in machine learning systems. The gorilla label was removed, and as of 2023 the system still could not recognise gorillas.

2014 · 12 years ago

Art history researchers applied a machine learning algorithm to fine art paintings

The reported study analysed paintings and may have revealed previously unrecognised influences among artists. It illustrated machine learning's spread into the humanities.

2014 · 12 years ago

Ian Goodfellow and others introduced generative adversarial networks

GANs pitted two networks against each other to produce realistic synthetic data. The architecture opened up generative modelling as a major machine learning capability.

2012 · 14 years ago

Vinod Khosla predicted machine learning would eliminate 80% of doctors' jobs

The Sun Microsystems co-founder forecast that automated machine learning medical diagnostic software would displace most physicians within two decades. The claim typified high expectations for medical machine learning.

2012 · 14 years ago

AlexNet marked the start of a 300,000-fold rise in training compute

OpenAI estimated the hardware compute used in the largest deep learning projects from AlexNet in 2012 to AlphaZero in 2017 and found a 300,000-fold increase, with a doubling time of 3.4 months. This scaling underpinned the deep learning era.

2010s · 16 years ago

Machine learning and hardware researchers developed more efficient methods for training deep neural networks

Advances in both algorithms and computer hardware enabled the training of networks with many layers of nonlinear hidden units. This combination made deep learning practical and allowed it to surpass earlier machine learning approaches.