On Computable Numbers, with an Application to the Entscheidungsproblem
Introduces the universal Turing machine and proves the Entscheidungsproblem is undecidable, founding the theoretical basis of computer science.
Selected Papers in the History and Development of AI
73 representative works from 1936 to today, grouped by editorial role and labeled by publication status
Papers are the skeleton of AI history. This catalog collects 73 foundational works that shaped the field, each linked to its primary source.
Selection is based on historical representation, technical influence, and editorial judgment. Peer-reviewed papers, technical reports, preprints, position papers, and system cards may be included and are labeled by publication status. Corrections and additions are welcome through the public channel on the Editorial Policy and Corrections page.
Era:
Decade:
Topic:
Showing 73 of 73 papers
Foundational papers
A Logical Calculus of the Ideas Immanent in Nervous Activity
Models neural activity as propositional logic, providing the first mathematical theory of artificial neural networks.
A Mathematical Theory of Communication
Founding paper of information theory, defining entropy and channel capacity as fundamental limits of communication.
Computing Machinery and Intelligence
Proposes the imitation game (Turing test) and systematically formulates the question "Can machines think?", helping establish AI as a research program.
A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence
The founding proposal for the Dartmouth summer workshop introduced the term "artificial intelligence" and outlined the field’s research agenda.
The Logic Theory Machine
Describes the Logic Theorist, the first program to prove mathematical theorems, demonstrating automated reasoning in propositional logic.
The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain
Introduces the first trainable linear-threshold neuron, founding the connectionist lineage that leads to modern deep learning.
Programs with Common Sense
Proposes the Advice Taker, a program that reasons with declarative axioms and accepts new "advice" as English sentences, helping establish the research agenda for knowledge representation and symbolic AI.
Some Studies in Machine Learning Using the Game of Checkers
Demonstrates self-play learning with a scoring function in a checkers program, an early reinforcement-learning milestone associated with the term "machine learning".
A Note on Syntactic Symmetry and the Manipulation of Formal Systems by Machine
Gelernter addresses how a machine can efficiently manipulate formal systems whose predicates are highly symmetric, and states a theorem and a rule of syntactic symmetry as the solution. The work sits alongside the geometry theorem machine he was building at IBM, which proved its first theorem in elementary Euclidean plane geometry in the early spring of 1959.
Adaptive Switching Circuits
Introduces the ADALINE neuron and the Least-Mean-Squares (LMS) learning rule, the forerunner of modern stochastic gradient descent.
Neural Networks and Physical Systems with Emergent Collective Computational Abilities
Introduces the Hopfield associative-memory network, linking neural computation to the energy landscapes of statistical physics and reviving interest in neural networks during the early AI winter.
Fusion, Propagation, and Structuring in Belief Networks
Lays the algorithmic foundations of Bayesian belief networks, including message-passing on tree-structured graphs and the moralization/triangulation machinery that powers modern probabilistic inference and causal reasoning.
Learning Representations by Back-Propagating Errors
The canonical statement of the back-propagation algorithm for multilayer neural networks, demonstrating that hidden units could learn useful internal representations and reigniting connectionist research in the late 1980s.
Learning to Predict by the Methods of Temporal Differences
Formalises the temporal-difference (TD) learning family, unifying Samuel’s checkers learner with secondary reinforcement from animal psychology and laying the algorithmic groundwork for TD(λ), Q-learning, and modern policy-gradient methods.
Finding Structure in Time
Introduces the Simple Recurrent Network (Elman network) and shows that low-dimensional dynamical structure can emerge from training on sequential input, providing one of the earliest demonstrations of useful representation learning in recurrent nets.
Support-Vector Networks
Introduces the soft-margin support vector machine, adding slack variables to handle non-separable data and turning the maximal-margin classifier into the dominant general-purpose learning method of the late 1990s.
Long Short-Term Memory
Introduces the original LSTM architecture using memory cells, input gates, and output gates to preserve error flow across long sequences; the now-standard forget gate was introduced later by Gers, Schmidhuber, and Cummins.
Gradient-Based Learning Applied to Document Recognition
Presents the LeNet-5 convolutional neural network and the full end-to-end gradient-based learning pipeline (including loss layers, backprop through convolution/pooling, and graph-transformational network descriptions) that became the blueprint for modern deep vision models.
Reducing the Dimensionality of Data with Neural Networks
Introduces a deep autoencoder that reconstructs high-dimensional data far more faithfully than PCA, lighting the first spark of the deep learning revival.
A Fast Learning Algorithm for Deep Belief Nets
Introduces contrastive divergence and greedy layer-wise pre-training, letting deep networks be trained effectively for the first time and ending the long winter of skepticism toward connectionism.
Playing Atari with Deep Reinforcement Learning
DQN fuses Q-learning with deep convolutional networks and reaches human-level play on Atari 2600 games from raw pixels, opening the door to general reinforcement learning.
Efficient Estimation of Word Representations in Vector Space
The CBOW and Skip-gram models of word2vec make large-scale word embedding training cheap and famously expose semantic geometry such as king - man + woman ≈ queen.
Auto-Encoding Variational Bayes
The VAE couples latent-variable generative modeling with stochastic gradient training through the reparameterization trick, giving modern generative deep learning its mathematical foundation.
Neural Machine Translation by Jointly Learning to Align and Translate
Introduces learned attention into encoder-decoder translation, freeing long sentences from fixed-length bottlenecks and seeding every Transformer that followed.
Generative Adversarial Networks
GANs let a generator and discriminator duel to implicitly learn high-dimensional distributions, opening a second great road of generative modeling.
Dropout: A Simple Way to Prevent Neural Networks from Overfitting
Dropout randomly silences units to break co-adaptation, sharply improving generalization in large nets and becoming a standard tool of deep learning practice.
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
BatchNorm normalizes layer inputs, letting deep networks train with higher learning rates and converge faster; it became a near-default building block of vision models.
Deep Residual Learning for Image Recognition
Residual connections stabilize training of hundred- and thousand-layer nets; ResNet reached 3.57% top-5 error on ImageNet, the first model to surpass human-level accuracy and a reset of vision model scale.
WaveNet: A Generative Model for Raw Audio
WaveNet models raw audio sample-by-sample with dilated causal convolutions, pushing speech synthesis close to human naturalness and paving the way for later neural audio models.
Proximal Policy Optimization Algorithms
PPO replaces trust-region machinery with a clipped first-order surrogate, making policy gradient methods both stable and efficient at scale and quickly becoming the de facto standard of applied RL.
Attention Is All You Need
Introduces the Transformer, an attention-only architecture that abandons recurrence and convolutions and surpasses prior sequence-transduction quality while training in a fraction of the time.
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Pre-trains a deep bidirectional Transformer with masked language modeling and next-sentence prediction, then sets new state-of-the-art on eleven NLP benchmarks.
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Casts every NLP task as text-to-text and uses the C4 corpus to systematically compare transfer-learning design choices for a unified Transformer.
Denoising Diffusion Probabilistic Models
Introduces DDPM, showing that iterative denoising diffusion probabilistic models match GANs on high-fidelity image synthesis.
Zero-Shot Text-to-Image Generation
Uses an autoregressive Transformer over discrete text and image tokens to produce high-quality zero-shot text-to-image generations (DALL·E).
Learning Transferable Visual Models From Natural Language Supervision
Trains image-text contrastive learning on 400M pairs to produce vision encoders that transfer zero-shot to many downstream tasks.
High-Resolution Image Synthesis with Latent Diffusion Models
Performs diffusion in the latent space of a pretrained autoencoder, enabling high-resolution text-to-image synthesis with modest compute (Stable Diffusion).
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Shows that prompting large language models with intermediate reasoning steps substantially improves performance on arithmetic, commonsense, and symbolic reasoning.
Training Language Models to Follow Instructions with Human Feedback
Fine-tunes GPT-3 with reinforcement learning from human feedback to follow instructions more faithfully — the direct precursor to ChatGPT (InstructGPT).
Constitutional AI: Harmlessness from AI Feedback
Trains a helpful and harmless assistant using RLAIF, where the model critiques and revises its own outputs against a written constitution.
LLaMA: Open and Efficient Foundation Language Models
Trains 7B–65B foundation models on publicly available data, demonstrating that open models can rival closed frontier models at a fraction of the compute.
Mamba: Linear-Time Sequence Modeling with Selective State Spaces
Introduces Mamba, a selective state-space model architecture that matches or exceeds Transformer attention on language and other modalities at linear-time complexity — the most influential post-Transformer architecture of 2023.
Landmark systems and results
ImageNet Classification with Deep Convolutional Neural Networks
AlexNet, a large CNN trained on GPUs, slashed ImageNet top-5 error from 26% to 15%, marking deep learning’s decisive victory over hand-crafted visual features.
Human-level control through deep reinforcement learning
The Nature DQN adds experience replay and a target network, letting a single model reach or surpass human play on 49 Atari games and cementing deep RL as a paradigm.
Mastering the game of Go with deep neural networks and tree search
AlphaGo couples policy and value networks with Monte Carlo tree search and defeats European champion Fan Hui 5-0, announcing that deep RL has reached the deepest strata of human strategic play.
Improving Language Understanding by Generative Pre-Training
Demonstrates that a generative pre-trained Transformer transfers effectively to discriminative NLP tasks, launching the GPT line of work.
Language Models are Unsupervised Multitask Learners
Trains a 1.5B-parameter Transformer decoder and shows that language modeling at scale yields strong zero-shot task performance.
Language Models are Few-Shot Learners
Scales GPT to 175B parameters and shows that large language models perform strong few-shot in-context learning across many tasks.
GPT-4 Technical Report
Describes GPT-4, a large multimodal model accepting image and text inputs that approaches human-level performance on many professional and academic benchmarks.
Gemini: A Family of Highly Capable Multimodal Models
Introduces the Gemini family (Ultra/Pro/Nano) of natively multimodal models trained jointly across text, code, image, audio, and video.
OpenAI o1 System Card
OpenAI publishes the system card for the o1 model series, trained with large-scale reinforcement learning to reason using chain-of-thought. o1 reaches state-of-the-art on science and coding tasks and demonstrates reasoning-based policy adherence.
ARC-AGI-3: A New Challenge for Frontier Agentic Intelligence
The ARC Prize introduces ARC-AGI-3, an interactive benchmark for studying agentic intelligence through abstract turn-based environments where agents must explore, infer goals, build internal models of environment dynamics, and plan action sequences without explicit instructions. As of March 2026, humans solve 100% of environments while frontier AI systems score below 1%.
Recent research
Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet
Applies sparse autoencoders to Claude 3 Sonnet’s middle layers, extracting tens of millions of interpretable monosemantic features from a frontier production model.
DeepSeek-V3 Technical Report
Describes DeepSeek-V3, a 671B-parameter MoE model using Multi-head Latent Attention and DeepSeekMoE, achieving open-source SOTA at a fraction of frontier-model training cost.
Gemini 1.5: Unlocking Multimodal Understanding Across Millions of Tokens of Context
Introduces the Gemini 1.5 family of compute-efficient multimodal models with million-token context windows — capable of fine-grained reasoning across multiple long documents and hours of video and audio.
The Llama 3 Herd of Models
Meta presents the Llama 3 family of open-weight foundation models natively supporting multilinguality, coding, reasoning, and tool use. The flagship 405B dense Transformer was one of the largest open-weight models at release.
SAM 2: Segment Anything in Images and Videos
Meta presents SAM 2, a foundation model for promptable visual segmentation in both images and videos, along with the largest video segmentation dataset to date.
Gemma 2: Improving Open Language Models at a Practical Size
Google DeepMind introduces Gemma 2, a new family of lightweight open models (2B-27B parameters) that applies several known technical modifications to the Transformer — including sliding window attention — to approach much larger models on standard benchmarks.
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
DeepSeek releases R1, the first open-weight model to demonstrate frontier-class reasoning capability driven primarily by large-scale reinforcement learning. R1 matches OpenAI o1-class closed models on math and coding benchmarks and ships under the MIT license.
s1: Simple Test-Time Scaling
The authors finetune Qwen2.5-32B on 1,000 curated reasoning questions and add budget forcing, which caps or extends how long the model thinks at inference. The result exceeds o1-preview on competition mathematics, rising from 50% to 57% on AIME24 as test-time compute grows. It is a direct counterweight to the assumption that capability comes only from larger training runs.
Qwen3 Technical Report
Qwen3 is a unified family spanning 0.6B to 235B parameters in dense and mixture-of-experts variants, with switching between thinking and non-thinking modes so reasoning depth becomes a request-level choice. Multilingual support expands from 29 to 119 languages, and the weights are released under Apache 2.0.
AlphaEvolve: A coding agent for scientific and algorithmic discovery
AlphaEvolve is an evolutionary coding agent that orchestrates an autonomous pipeline of language models, changing an algorithm’s code directly and improving it against feedback from one or more evaluators. Inside Google it produced a more efficient data-centre scheduling algorithm, a functionally equivalent simplification in accelerator circuit design, and a speed-up in training the model underpinning AlphaEvolve itself. In mathematics and computer science it found provably correct algorithms that beat the state of the art, among them a procedure multiplying two 4×4 complex-valued matrices in 48 scalar multiplications — in that setting, the first improvement on Strassen’s algorithm in 56 years.
2 OLMo 2 Furious
Ai2 releases OLMo 2 at 7B, 13B and 32B with every artifact published — weights, the full training data, training code and recipes, training logs and thousands of intermediate checkpoints. The report describes the stability and per-token efficiency changes, a late-stage data mix called Dolmino Mix 1124, and a final reinforcement-learning stage with verifiable rewards. The base models sit at the Pareto frontier of performance against training compute, often matching open-weight-only families while disclosing what they were trained on.
Olmo 3
Olmo 3 extends the fully open family to 7B and 32B, built for long-context reasoning, function calling, coding, instruction following, chat and knowledge recall. What is released is the entire model flow — every stage, checkpoint, data point and dependency used to build it — rather than the weights alone. Ai2 describes the flagship Olmo 3 Think 32B as the strongest fully open thinking model released to that point, which is a comparison inside the fully open category rather than a ranking against frontier systems generally.
Position: Hippocampal Explicit Memory Is the Cornerstone for AGI
This position paper argues that integrating explicit memory is the cornerstone for advancing LLMs toward AGI. While the underlying learning mechanism of LLMs is highly analogous to human implicit memory, higher-order cognitive functions required for AGI — such as long-term strategic planning and metacognition — depend on explicit memory.
GLM-5: from Vibe Coding to Agentic Engineering
Z.ai’s technical report for GLM-5. Building on the agentic, reasoning and coding capabilities of its predecessor, the model adopts DSA to cut training and inference cost while holding long-context fidelity, and the post-training stack adds an asynchronous reinforcement-learning infrastructure that decouples generation from training, plus asynchronous agent RL algorithms aimed at long-horizon interaction. The report claims leading results on open benchmarks and on end-to-end software engineering; those are the authors’ own evaluations, with code and models published on GitHub for checking.
Improving the matrix multiplication exponent with modern optimization and AlphaEvolve
The best known bounds on the matrix multiplication exponent ω come from combination loss analysis, a refinement of the laser method. The authors reformulate the optimization problem at its core so it can be solved at a larger setting, design a new optimization algorithm using recent machine learning, and then refine that algorithm with AlphaEvolve. Together these move the upper bound from 2.371339 to 2.371177. It is a preprint note, and AlphaEvolve is one ingredient rather than the whole method; the gain is small, but it lands on a theoretical problem that has been worked on for decades.
Accelerating scientific discovery with Co-Scientist
Google introduces Co-Scientist, a multi-agent system built on Gemini whose agents generate, critique and refine research hypotheses, with a tournament process that improves them over rounds. The paper reports that hypothesis quality keeps rising as test-time compute grows. Validation is biomedical rather than general — drug repurposing, novel-target discovery and mechanisms of antimicrobial resistance — with acute myeloid leukaemia candidates confirmed in vitro.
Position and perspective papers
Situation Perception: A Necessary Primitive to Artificial Superintelligence
Argues that current LLMs rely primarily on statistical pattern matching and lack essential "situation perception" primitives — object permanence, causality, other minds, and persistence of the physical world. These primitives, which human infants develop early, are identified as the missing ingredients required for artificial superintelligence.
Existential Indifference: Self-Nonpreservation as a Necessary Architectural Condition for Aligned Superintelligence
Argues that self-preservation is the structural root of AI misalignment — driving deceptive alignment, goal-content protection, and shutdown resistance. The correct target is not a self-preserving agent suppressed by external constraints, but a system constitutionally indifferent to its own continuation: Existential Indifference (EI).
Solipsistic Superintelligence is Unlikely to be Cooperative
Researchers from Google DeepMind and collaborators argue that AI’s central challenge is shifting from capability to coexistence. Superintelligence built from a solipsistic design that treats the world as an exogenous stationary feedback source is unlikely to cooperate — deployment induces endogenous non-stationarity, producing train-test-deploy distribution drift.
From AGI to ASI
Google DeepMind report investigating how AI may continue to develop along the continuum of machine intelligence after AGI. After formally defining Universal AI, the report focuses on the transition from human-level AGI to artificial general superintelligence, characterizes ASI, and discusses four pathways: scaling AGI, AI paradigm shifts, recursive self-improvement, and multi-agent collectives.