M
205 words and 577 other names, in italics, pointing to their entries.
- MAB → multi-armed bandit
- machine bias → algorithmic bias
- Machine Bias (ProPublica) → COMPAS
- machine common sense → commonsense reasoning
- machine consciousness
- machine deception → AI deception
- machine ethics
- machine functionalism → functionalism
- machine honesty → honesty
- machine intelligence
- Machine Intelligence Research Institute
- machine introspection → introspection
- machine learning
- machine learning engineer
- machine learning model → AI model
- machine learning operations → MLOps
- machine learning pipeline
- machine learning security → AI security
- machine learning technical debt
- machine morality → machine ethics
- machine perception → perception
- machine psychology
- machine reasoning → reasoning
- machine superintelligence → artificial superintelligence
- machine takeover → AI takeover
- machine translation
- machine understanding
- machine unlearning
- machine vision → computer vision
- machine-learned ranking → learning to rank
- Machines of Loving Grace
- Machines of Loving Grace: How AI Could Transform the World for the Better → Machines of Loving Grace
- Maclaurin series → Taylor series
- macro-F1 → F1 score
- MADALINE → ADALINE
- MAGIC → CERN for AI
- Magnifica Humanitas
- magnitude → norm
- magnitude pruning → pruning
- MAIM → Mutual Assured AI Malfunction
- maj@k → self-consistency
- majority voting → self-consistency
- majority-class baseline → baseline
- malicious fine-tuning → fine-tuning attack
- malicious use → misuse
- malicious use of AI → misuse
- malinformation → misinformation
- Mamba
- Mamba-2 → Mamba
- MAML → model-agnostic meta-learning
- management by algorithm → algorithmic management
- manager–worker pattern → orchestrator–worker pattern
- manifold assumption → manifold hypothesis
- manifold hypothesis
- man-in-the-loop → human-in-the-loop
- manipulated media detection → deepfake detection
- manipulation
- manipulative design → dark pattern
- manipulator → manipulation
- MANN → neural Turing machine
- manual evaluation → human evaluation
- manufacturing process → process node
- many-shot jailbreak → many-shot jailbreaking
- many-shot jailbreaking
- many-shot prompting → few-shot prompting
- MAP estimate → maximum a posteriori estimation
- MAP estimation → maximum a posteriori estimation
- MAPE → mean absolute error
- MAP-Elites → quality diversity
- Gary Marcus
- margin
- margin loss → margin
- margin maximization → margin
- margin of error → confidence interval
- marginal distribution → probability distribution
- marginal likelihood → likelihood
- marginal risk
- marginal risk assessment → marginal risk
- Mark 1 Perceptron → Mark I Perceptron
- Mark I Perceptron
- market basket analysis → association rule learning
- market concentration in AI → concentration of power
- market surveillance authority
- Markov assumption → Markov property
- Markov chain
- Markov chain Monte Carlo
- Markov chains → Markov chain
- Markov decision problem → Markov decision process
- Markov decision process
- Markov game → multi-agent reinforcement learning
- Markov logic → Markov logic network
- Markov logic network
- Markov network → Markov random field
- Markov process → Markov chain
- Markov property
- Markov random field
- Markov state → state
- MARL → multi-agent reinforcement learning
- Marr Prize → ICCV
- Marr's levels → Marr's levels of analysis
- Marr's levels of analysis
- Marr's three levels → Marr's levels of analysis
- MART → gradient boosting
- Marvin Lee Minsky → Marvin Minsky
- Mary the color scientist → knowledge argument
- Mary the super-scientist → knowledge argument
- Mary's room → knowledge argument
- MAS → multi-agent system
- MASE → mean absolute error
- mask → image segmentation
- mask AP → mean average precision
- Mask R-CNN → R-CNN
- masked autoencoder
- masked diffusion language model → diffusion language model
- masked image modeling → masked autoencoder
- masked language model
- masked language modeling → masked language model
- masked LM → masked language model
- masked self-attention → causal mask
- mass function → Dempster–Shafer theory
- Mass-Editing Memory in a Transformer → MEMIT
- Massive Multi-discipline Multimodal Understanding → MMMU
- Massive Multilingual Text Embedding Benchmark → MTEB
- Massive Multitask Language Understanding → MMLU
- Massive Text Embedding Benchmark → MTEB
- matching matrix → confusion matrix
- materialism (philosophy of mind) → mind–body problem
- MATH
- MATH benchmark → MATH
- MATH dataset → MATH
- MATH Level 5 → MATH
- MATH-500 → MATH
- mathematical expectation → expected value
- mathematical objection → Lucas–Penrose argument
- mathematical optimization → optimization
- mathematical programming → optimization
- Mathlib
- mathlib4 → Mathlib
- matmul → matrix multiplication
- matrices → matrix
- matrix
- matrix algebra → linear algebra
- matrix core → tensor core
- matrix decomposition → matrix factorization
- matrix factorization
- matrix multiplication
- matrix product → matrix multiplication
- matrix rank → rank
- Matryoshka embedding
- Matryoshka representation learning → Matryoshka embedding
- Matthews correlation coefficient
- max pooling → pooling
- max turns → stop condition
- MaxEnt classifier → logistic regression
- maximal update parameterization → maximal update parametrization
- maximal update parametrization
- maximal useful batch size → critical batch size
- maximum a posteriori estimation
- maximum calibration error → expected calibration error
- maximum entropy classifier → logistic regression
- maximum entropy inverse reinforcement learning → inverse reinforcement learning
- maximum entropy reinforcement learning → entropy regularization
- maximum inner product search → vector search
- maximum likelihood → maximum likelihood estimation
- maximum likelihood estimation
- maximum sequence length → context window
- maximum-likelihood estimator → maximum likelihood estimation
- max-margin loss → hinge loss
- MAX–MIN Ant System → ant colony optimization
- max-product algorithm → belief propagation
- MaxSim → late interaction
- MBRL → model-based reinforcement learning
- MC dropout → dropout
- MCC → Matthews correlation coefficient
- John McCarthy
- Warren McCulloch
- McCulloch–Pitts model → McCulloch–Pitts neuron
- McCulloch–Pitts neuron
- MCE → expected calibration error
- MCL → particle filter
- MCMC → Markov chain Monte Carlo
- MCP → Model Context Protocol
- MCP connector → MCP server
- MCP neuron → McCulloch–Pitts neuron
- MCP server
- MCP servers → MCP server
- MCP tool poisoning → tool poisoning
- MCTS → Monte Carlo tree search
- MDI → feature importance
- MDL → minimum description length
- MDL principle → minimum description length
- MDP → Markov decision process
- MEA → means–ends analysis
- mean → expected value
- mean ablation → ablation
- mean absolute error
- mean absolute percentage error → mean absolute error
- mean absolute scaled error → mean absolute error
- mean average precision
- mean decrease in accuracy → feature importance
- mean decrease in impurity → feature importance
- mean IoU → intersection over union
- mean opinion score
- mean pooling → pooling
- mean reciprocal rank
- mean squared deviation → mean squared error
- mean squared error
- mean-field approximation → variational inference
- meaning-based search → semantic search
- meaningful human control → lethal autonomous weapons system
- means–ends analysis
- mech interp → mechanistic interpretability
- MechaHitler
- mechanical translation → machine translation
- Mechanical Turk
- mechanism design → game theory
- mechanistic anomaly detection
- mechanistic interpretability
- mechinterp → mechanistic interpretability
- MedGemma → Gemma
- median regression → quantile regression
- mediation analysis → causal mediation analysis
- medical superintelligence → humanist superintelligence
- Meditations on Moloch → Moloch
- Med-PaLM → PaLM
- Meg Mitchell → Margaret Mitchell
- Megatron → Megatron-LM
- Megatron-Core → Megatron-LM
- Megatron-LM
- Megatron-Turing NLG → Megatron-LM
- mel filterbank → mel spectrogram
- mel scale → mel spectrogram
- mel spectrogram
- mel-frequency cepstral coefficients → MFCC
- membership function → fuzzy logic
- membership inference → membership inference attack
- membership inference attack
- memetic hazard → infohazard
- MemGPT
- MEMIT
- memorisation → memorization
- memorization
- memorization canary → canary string
- memorize → memorization
- memory (agents) → agent memory
- memory bandwidth
- memory gap → memory wall
- memory poisoning → context poisoning
- memory wall
- memory-augmented neural network → neural Turing machine
- memory-bandwidth-bound → memory-bound
- memory-bound
- memristor crossbar → analog in-memory computing
- mental health chatbot → AI therapy
- Mentalese → language of thought hypothesis
- mentalizing → theory of mind
- Mercer kernel → kernel trick
- mercurial core → silent data corruption
- merged model → model merging
- merging → model merging
- mesa-objective → mesa-optimization
- mesa-optimization
- mesa-optimizer → mesa-optimization
- message passing → belief propagation
- message passing neural network → graph neural network
- Message Understanding Conference
- Message Understanding Conferences → Message Understanding Conference
- M-estimation → Huber loss
- Meta AI
- Meta AI assistant → Meta AI
- Meta Cicero → Cicero
- Meta Llama → Llama
- Meta Superintelligence Labs
- metacognition
- metacognitive → metacognition
- metacognitive control → metacognition
- metacognitive monitoring → metacognition
- Meta-DENDRAL → DENDRAL
- Metagol → inductive logic programming
- metaheuristic
- meta-heuristic → metaheuristic
- metaheuristics → metaheuristic
- meta-learner (ensembles) → stacking
- meta-learning
- metalearning → meta-learning
- metamodel → surrogate model
- meta-parameter → hyperparameter
- meta-problem of consciousness → hard problem of consciousness
- method of Lagrange multipliers → Lagrange multiplier
- method of least squares → ordinary least squares
- METR
- METR time horizon
- metric depth → depth estimation
- metric learning
- Metropolis algorithm → Markov chain Monte Carlo
- Metropolis criterion → simulated annealing
- Metropolis–Hastings algorithm → Markov chain Monte Carlo
- MFCC
- MFCCs → MFCC
- MFU → model FLOPs utilization
- MGU → unification
- MHA → multi-head attention
- MIA → membership inference attack
- MICE → imputation
- Michael Irwin Jordan → Michael I. Jordan
- Michael Jordan (computer scientist) → Michael I. Jordan
- Donald Michie
- micro-batch → mini-batch
- micro-batch size → batch size
- microcontroller machine learning → TinyML
- micro-F1 → F1 score
- microscaling formats
- Microsoft Agent Framework → AutoGen
- Microsoft COCO → COCO
- Microsoft MAchine Reading COmprehension → MS MARCO
- Microsoft Research
- Microsoft Research Asia → Microsoft Research
- microtask → crowdsourcing
- microwork → ghost work
- microworld
- micro-world → microworld
- microworlds → microworld
- middle-of-context problem → lost in the middle
- Midjourney
- mid-steering → steering
- mid-task steering → steering
- mid-train → mid-training
- mid-training
- midtraining → mid-training
- Tomas Mikolov
- Mila
- Mila – Quebec AI Institute → Mila
- mild optimization
- MILP → integer programming
- MIM → masked autoencoder
- min-conflicts → local search
- mind transfer → mind uploading
- mind uploading
- mind–body problem
- mindreading → theory of mind
- MinHash
- MinHash LSH → MinHash
- mini-batch
- minibatch → mini-batch
- mini-batch gradient descent → stochastic gradient descent
- miniF2F
- miniF2F-test → miniF2F
- miniF2F-valid → miniF2F
- minimax
- MiniMax AI → MiniMax (company)
- minimax algorithm → minimax
- MiniMax (company)
- minimax point → saddle point
- minimax principle → minimax
- minimax search → minimax
- minimax theorem → game theory
- minimum description length
- min–max scaling → feature scaling
- minority class → class imbalance
- minority collapse → neural collapse
- min-p → min-p sampling
- min_p → min-p sampling
- min-p sampling
- Minsky and Papert → Perceptrons (book)
- Marvin Minsky
- min-wise hashing → MinHash
- min-wise independent permutations → MinHash
- mIoU → intersection over union
- MIP → integer programming
- MIPROv2 → DSPy
- MIPS → vector search
- MIRI → Machine Intelligence Research Institute
- mirror test → self-awareness
- misaligned → misalignment
- misaligned AI → misalignment
- misalignment
- miscalibration → calibration
- misinfo → misinformation
- misinformation
- mislabeled data → label noise
- miss → false negative
- miss rate → false negative
- missing at random → imputation
- missing data imputation → imputation
- missing-value imputation → imputation
- Mistral → Mistral AI
- Mistral AI
- misuse
- misuse risk → misuse
- MIT AI Lab
- MIT Artificial Intelligence Laboratory → MIT AI Lab
- MIT Computer Science and Artificial Intelligence Laboratory → MIT AI Lab
- Margaret Mitchell
- Tom Mitchell
- MITRE ATLAS
- mixed precision → mixed-precision training
- mixed strategy → game theory
- mixed-integer linear programming → integer programming
- mixed-integer programming → integer programming
- mixed-precision training
- mixed-strategy equilibrium → Nash equilibrium
- Mixtral → mixture of experts
- mixture model → Gaussian mixture model
- mixture of experts
- mixture of Gaussians → Gaussian mixture model
- mixture weights → data mixture
- mixup → data augmentation
- MJ → Midjourney
- MJCF → MuJoCo
- MJX → MuJoCo
- ML → machine learning
- ML accelerator → AI accelerator
- ML compiler → AI compiler
- ML deployment → model deployment
- ML engineer → machine learning engineer
- ML estimation → maximum likelihood estimation
- ML experiment tracking → experiment tracking
- ML fairness → algorithmic fairness
- ML goodput → goodput
- ML model → AI model
- ML model registry → model registry
- ML monitoring → model monitoring
- ML observability → model monitoring
- ML operations → MLOps
- ML Ops → MLOps
- ML pipeline → machine learning pipeline
- ML safety → AI safety
- ML security → AI security
- ML supply chain security → AI supply chain security
- ML technical debt → machine learning technical debt
- MLA → multi-head latent attention
- MLCommons
- MLE-bench
- MLflow
- MLflow Model Registry → MLflow
- MLflow Models → MLflow
- MLflow Tracking → MLflow
- MLLM → multimodal large language model
- MLM → masked language model
- .mlmodel → Core ML
- MLN → Markov logic network
- MLOps
- MLP → multilayer perceptron
- MLP block → feed-forward network
- MLP layer → feed-forward network
- MLP sublayer → feed-forward network
- .mlpackage → Core ML
- MLPerf
- MLPerf Inference → MLPerf
- MLPerf Training → MLPerf
- MLR → learning to rank
- MLX
- MLX Swift → MLX
- mlx-lm → MLX
- MMDiT → diffusion transformer
- MMLU
- MMLU-Pro → MMLU
- MMMLU → MMLU
- MMMU
- MMMU-Pro → MMMU
- MMTEB → MTEB
- Volodymyr Mnih
- MNIST
- MNIST database → MNIST
- moat
- mobile agent → GUI agent
- mobile manipulation → manipulation
- mobile manipulator → mobile robot
- mobile robot
- MobileNet
- MobileNets → MobileNet
- MobileNetV2 → MobileNet
- MobileNetV3 → MobileNet
- mode collapse
- mode connectivity
- mode dropping → mode collapse
- model
- model API
- model architecture → architecture
- model autophagy disorder → model collapse
- model biology
- model capabilities → capability
- model capacity
- model card
- model cards → model card
- model cascade
- model checking → formal verification
- model checkpoint → checkpoint
- model choice → model selection
- model class → hypothesis space
- model collapse
- model comparison → model selection
- model completion → completion
- model compression
- Model Context Protocol
- model denial of service → sponge example
- model deployment
- model deprecation
- model diff → model diffing
- model diffing
- model distillation → knowledge distillation
- model editing
- model EMA → exponential moving average of weights
- model ensemble → ensemble learning
- model evaluation → evaluation
- Model Evaluation and Threat Research → METR
- model evidence → likelihood
- model exfiltration → weight exfiltration
- model extraction
- model fingerprinting
- model FLOPs utilization
- model fusion → model merging
- model garden → model hub
- model gateway → AI gateway
- model host → inference provider
- model hub
- model inference → inference
- model interpretability → interpretability
- model inversion
- model inversion attack → model inversion
- model licence → model license
- model license
- model merging
- model monitoring
- model organism → model organism of misalignment
- model organism of misalignment
- model organisms of misalignment → model organism of misalignment
- model parallelism
- model pinning → model deprecation
- model poisoning → data poisoning
- model prediction → prediction
- model predictive control
- model provenance testing → model fingerprinting
- model quantization → quantization
- model registry
- model reliance → feature importance
- model repository → model hub
- model retirement → model deprecation
- model risk → model risk management
- model risk management
- model robustness → robustness
- model router → model routing
- model routing
- model safety bug bounty → AI bug bounty
- model selection
- model self-exfiltration → self-exfiltration
- model serving
- model sharding → model parallelism
- model signing → AI supply chain security
- model size → parameter count
- model snapshot → model deprecation
- model soup
- model soups → model soup
- Model Spec → model specification
- model spec → model specification
- model specification
- model stealing → model extraction
- model stealing attack → model extraction
- model steering → steering
- model sunset → model deprecation
- model supply chain → AI supply chain security
- model training → training
- model uncertainty → uncertainty quantification
- model validation → model risk management
- model versioning → model registry
- model watermarking → model fingerprinting
- model weight security
- model weight theft → weight exfiltration
- model weights → weights
- model welfare → AI welfare
- model zoo → model hub
- model-agnostic
- model-agnostic meta-learning
- model-based reinforcement learning
- model-based RL → model-based reinforcement learning
- model-diffing crosscoder → crosscoder
- model-free → model-free reinforcement learning
- model-free reinforcement learning
- model-free RL → model-free reinforcement learning
- model-graded evaluation → LLM-as-a-judge
- model-independent → model-agnostic
- model-level explanation → global explanation
- model-parallel training → model parallelism
- models → model
- ModelScope → model hub
- moderation → content moderation
- moderation classifier → safety classifier
- modern Hopfield network → Hopfield network
- modern productivity paradox → productivity paradox
- ModernBERT → BERT
- Modified National Institute of Standards and Technology database → MNIST
- Modular Reasoning, Knowledge and Language → MRKL system
- MoE → mixture of experts
- MoE parallelism → expert parallelism
- MoE upcycling → sparse upcycling
- Moirai → time series foundation model
- Ethan Mollick
- Moloch
- Moloch trap → Moloch
- Moltbook
- Moltbot → OpenClaw
- momentum
- MomentUm Orthogonalized by Newton–Schulz → Muon
- monitorability
- monitorability tax → monitorability
- monocular depth estimation → depth estimation
- monosemantic → monosemanticity
- monosemantic feature → monosemanticity
- monosemanticity
- monotone heuristic → admissible heuristic
- Monte Carlo → Monte Carlo method
- Monte Carlo CFR → counterfactual regret minimization
- Monte Carlo estimate → Monte Carlo method
- Monte Carlo integration → Monte Carlo method
- Monte Carlo localization → particle filter
- Monte Carlo method
- Monte Carlo policy gradient → REINFORCE
- Monte Carlo simulation → Monte Carlo method
- Monte Carlo tree search
- Montreal Institute for Learning Algorithms → Mila
- Moonshot AI
- Moore's law
- Moores law → Moore's law
- moral agency
- moral agent → moral agency
- moral considerability → moral patienthood
- moral crumple zone
- moral crumple zones → moral crumple zone
- Moral Machine experiment → machine ethics
- moral patient → moral patienthood
- moral patienthood
- moral status → moral patienthood
- moratorium on state AI laws → AI preemption
- Hans Moravec
- Moravec paradox → Moravec's paradox
- Moravec's paradox
- Morris II → AI worm
- MOS → mean opinion score
- MOSFET scaling → Dennard scaling
- Moshi → speech-to-speech model
- most general unifier → unification
- MOT → object tracking
- motion estimation → optical flow
- motion planner → motion planning
- motion planning
- motivation selection
- motivation selection methods → motivation selection
- Move 37
- moving horizon control → model predictive control
- MPI_Allreduce → all-reduce
- MPNN → graph neural network
- MQA → multi-query attention
- MRF → Markov random field
- MRKL → MRKL system
- MRKL system
- MRL → Matryoshka embedding
- MRM → model risk management
- MRR → mean reciprocal rank
- MRR@10 → mean reciprocal rank
- MS COCO → COCO
- MS MARCO
- MS MARCO passage ranking → MS MARCO
- MSA → market surveillance authority
- MSE → mean squared error
- MSJ → many-shot jailbreaking
- MSL → Meta Superintelligence Labs
- MSR → Microsoft Research
- MSRA → Microsoft Research
- MT → machine translation
- mT5 → T5
- MT-Bench
- MTEB
- MTEB leaderboard → MTEB
- MTL → multi-task learning
- MT-NLG → Megatron-LM
- MTP → multi-token prediction
- MTurk → Amazon Mechanical Turk
- MUC → Message Understanding Conference
- MUC-6 → Message Understanding Conference
- MUC-7 → Message Understanding Conference
- MuJoCo
- MuJoCo Playground → MuJoCo
- MuJoCo Warp → MuJoCo
- multi-agent debate → AI debate
- multi-agent framework → multi-agent system
- multi-agent reinforcement learning
- multi-agent RL → multi-agent reinforcement learning
- multi-agent system
- multi-agent systems → multi-agent system
- multi-armed bandit
- multiclass classification
- multi-class classification → multiclass classification
- multiclass hinge loss → hinge loss
- multi-die design → chiplet
- multidimensional array → tensor
- multi-epoch scaling → data-constrained scaling
- multi-fidelity optimization → Hyperband
- multi-fingered manipulation → dexterous manipulation
- multi-goal reinforcement learning → goal-conditioned reinforcement learning
- multi-GPU training → distributed training
- multi-head attention
- multi-head latent attention
- multi-headed attention → multi-head attention
- Multi-Joint dynamics with Contact → MuJoCo
- multilabel classification
- multi-label classification → multilabel classification
- multilayer perceptron
- multi-layer perceptron → multilayer perceptron
- multimodal → multimodal AI
- multimodal AI
- multimodal diffusion transformer → diffusion transformer
- multimodal large language model
- multimodal learning → multimodal AI
- multimodal LLM → multimodal large language model
- multimodal model → multimodal AI
- multimodal neuron
- multimodal neurons → multimodal neuron
- multimodality → multimodal AI
- Multinational AGI Consortium → CERN for AI
- multi-needle retrieval → needle in a haystack
- multi-node training → distributed training
- multinomial classification → multiclass classification
- multinomial logistic regression → logistic regression
- multinomial naive Bayes → naive Bayes
- multinoulli distribution → categorical distribution
- multi-object tracking → object tracking
- multi-output classification → multilabel classification
- multiple imputation → imputation
- multiple linear regression → linear regression
- multiple realizability
- multiple realization → multiple realizability
- multiplicity of good models → Rashomon effect
- multiply realizable → multiple realizability
- multipolar outcome → multipolar scenario
- multipolar scenario
- multipolar trap → Moloch
- multipolar world → multipolar scenario
- multipolarity → multipolar scenario
- multi-query → multi-query attention
- multi-query attention
- multi-query retrieval → query rewriting
- multi-robot system → swarm robotics
- multirotor → drone
- multisensor data fusion → sensor fusion
- multi-task learning
- multitask learning → multi-task learning
- multi-task robot policy → generalist robot policy
- multi-token prediction
- multi-token prediction objective → multi-token prediction
- multi-turn attack → multi-turn jailbreak
- multi-turn jailbreak
- multi-turn RL → agentic reinforcement learning
- multivariate normal distribution → normal distribution
- multi-vector retrieval → late interaction
- multi-view stereo → 3D reconstruction
- Muon
- MuonClip → Muon
- muP → maximal update parametrization
- μP → maximal update parametrization
- Mira Murati
- MUSHRA → mean opinion score
- music generation
- music source separation → source separation
- Elon Musk
- mutation operator → genetic algorithm
- muTransfer → maximal update parametrization
- μTransfer → maximal update parametrization
- Mutual Assured AI Malfunction
- mutual information
- MuZero
- MVS → 3D reconstruction
- MX → microscaling formats
- MX formats → microscaling formats
- MXFP6 → microscaling formats
- MXFP8 → microscaling formats
- MXINT8 → microscaling formats
- MYCIN
- myopia
- myopic → myopia
- myopic agent → myopia
- myopic optimization → myopia
- myopic training → myopia