D
159 words and 490 other names, in italics, pointing to their entries.
- D. Raj Reddy → Raj Reddy
- DAA → direct alignment algorithm
- Dabbala Rajagopal Reddy → Raj Reddy
- d/acc → differential technological development
- DAE → denoising autoencoder
- DAgger
- DAIR → Distributed AI Research Institute
- DALL·E
- DALLE → DALL·E
- DALL·E 2 → DALL·E
- DALL·E 3 → DALL·E
- DAN
- Dan Dennett → Daniel Dennett
- DAN jailbreak → DAN
- DAN prompt → DAN
- dangerous capabilities evaluation → dangerous capability evaluation
- dangerous capability eval → dangerous capability evaluation
- dangerous capability evaluation
- dangerously skip permissions → YOLO mode
- Daniel C. Dennett → Daniel Dennett
- DAPT → continued pretraining
- dark factory
- dark launch → shadow deployment
- dark pattern
- dark patterns → dark pattern
- DARPA
- DARPA Grand Challenge
- DARPA Urban Challenge → DARPA Grand Challenge
- Dartmouth conference → Dartmouth workshop
- Dartmouth proposal → Dartmouth workshop
- Dartmouth Summer Research Project on Artificial Intelligence → Dartmouth workshop
- Dartmouth workshop
- DARTS → neural architecture search
- Darwin Gödel Machine
- DAS → distributed alignment search
- data
- data annotation → annotation
- data attribution
- data augmentation
- data bottleneck → data wall
- data card → datasheet for datasets
- data center in space → orbital data center
- data center infrastructure efficiency → power usage effectiveness
- data cleaning
- data cleansing → data cleaning
- data colonialism → AI colonialism
- data contamination
- data curation pipeline → data pipeline
- data drift
- data efficiency → sample efficiency
- data engine → data flywheel
- data exhaustion → data wall
- data extraction attack → training data extraction
- data filtering → quality filtering
- data flywheel
- data fusion → sensor fusion
- data labeling
- data labelling → data labeling
- data leakage
- data license → data licensing
- data licensing
- data lineage → data provenance
- data manifold → manifold hypothesis
- data mining
- data mix → data mixture
- data mixing → data mixture
- data mixture
- data network effect → data flywheel
- data normalization → feature scaling
- data parallelism
- data pipeline
- data point → example
- data poisoning
- data privacy → privacy
- data processing pipeline → data pipeline
- data programming → weak supervision
- data protection
- data protection law → data protection
- data provenance
- data quality
- data science
- data scientist → data science
- data scraping → web scraping
- data scrubbing → data cleaning
- data set → dataset
- data shift → distribution shift
- data version control → data versioning
- data versioning
- data wall
- data work → ghost work
- data-centric AI
- data-centric approach → data-centric AI
- data-centric machine learning → data-centric AI
- data-constrained scaling
- data-constrained scaling laws → data-constrained scaling
- data-driven inference → forward chaining
- data-driven reasoning → forward chaining
- dataflow graph → computational graph
- DataFrame → pandas
- Datalog
- data-parallel training → data parallelism
- data-protection law → data protection
- dataset
- Dataset Aggregation → DAgger
- dataset card → datasheet for datasets
- dataset license → data licensing
- dataset poisoning → data poisoning
- dataset provenance → data provenance
- dataset shift → distribution shift
- dataset versioning → data versioning
- datasheet → datasheet for datasets
- datasheet for datasets
- datasheets for datasets → datasheet for datasets
- data-to-model ratio → tokens per parameter
- data-to-text generation → natural language generation
- dated snapshot → model deprecation
- datum → data
- David E. Rumelhart → David Rumelhart
- David Everett Rumelhart → David Rumelhart
- David J. Chalmers → David Chalmers
- davinci → GPT-3
- DBN → deep belief network
- DBSCAN
- DC eval → dangerous capability evaluation
- DCASE → sound event detection
- DCG → normalized discounted cumulative gain
- DCGAN → generative adversarial network
- DCiE → power usage effectiveness
- DDIM → denoising diffusion implicit model
- DDIM inversion → denoising diffusion implicit model
- DDIM sampler → denoising diffusion implicit model
- DDP → data parallelism
- DDPG → deep deterministic policy gradient
- DDPM → denoising diffusion probabilistic model
- DDPMs → denoising diffusion probabilistic model
- DDR → DRAM
- de facto Brussels effect → Brussels effect
- de jure Brussels effect → Brussels effect
- dead grandma trick → grandma exploit
- dead internet → dead internet theory
- dead internet theory
- dead latent → sparse autoencoder
- dead neurons → dying ReLU
- dead reckoning → odometry
- dead ReLU → dying ReLU
- deadly triad
- Jeff Dean
- debate → AI debate
- DeBERTa → BERT
- decel
- deceleration → decel
- decelerationist → decel
- decels → decel
- decentralised training → decentralized training
- decentralized SGD → decentralized training
- decentralized training
- deception detection → AI lie detection
- deceptive AI → AI deception
- deceptive alignment
- deceptive design → dark pattern
- deceptive misalignment → deceptive alignment
- deceptive pattern → dark pattern
- deceptively aligned → deceptive alignment
- deceptively aligned model → deceptive alignment
- decision boundary
- decision model
- decision region → decision boundary
- decision rule → policy
- decision stump → decision tree
- decision surface → decision boundary
- decision theory
- decision threshold
- decision transformer
- decision tree
- decision tree learning → decision tree
- decision-theoretic → decision theory
- decisive strategic advantage
- Declarative Self-improving Python → DSPy
- decode → decode phase
- decode phase
- decode step → decode phase
- decoder → encoder–decoder architecture
- decoder block → transformer block
- decoder-only architecture → decoder-only transformer
- decoder-only model → decoder-only transformer
- decoder-only transformer
- decoding
- decoding algorithm → decoding
- decoding phase → decode phase
- decoding strategy → decoding
- decolonial AI → AI colonialism
- decompose → task decomposition
- decomposition prompting → least-to-most prompting
- decontamination → data contamination
- deconvnet visualization → guided backpropagation
- deconvolution → transposed convolution
- decoupled weight decay → AdamW
- deduction → deductive reasoning
- deductive database → Datalog
- deductive inference → deductive reasoning
- deductive logic → deductive reasoning
- deductive program synthesis → program synthesis
- deductive reasoning
- deductive verification → formal verification
- dedup → deduplication
- dedupe → deduplication
- deduplication
- deep belief net → deep belief network
- deep belief network
- Deep Blue
- Deep Blue (malaise)
- Deep CFR → counterfactual regret minimization
- deep deterministic policy gradient
- deep double descent → double descent
- Deep Dream → DeepDream
- deep ensemble → ensemble learning
- deep ensembles → uncertainty quantification
- deep fake → deepfake
- deep generative model → generative model
- deep leakage from gradients → gradient inversion attack
- deep learning
- deep learning accelerator → AI accelerator
- deep learning boom → deep learning revolution
- deep learning compiler → AI compiler
- Deep Learning Important FeaTures → DeepLIFT
- deep learning is hitting a wall → scaling wall
- deep learning renaissance → deep learning revolution
- deep learning revolution
- deep metric learning → metric learning
- deep neural network → deep learning
- deep Q-learning → deep Q-network
- deep Q-network
- deep reinforcement learning
- deep research
- deep research agent → deep research
- deep residual network → ResNet
- deep RL → deep reinforcement learning
- Deep Sets → invariance
- deep synthesis
- deep synthesis technology → deep synthesis
- deep Taylor decomposition → layer-wise relevance propagation
- Deep Thought
- DeepDream
- Deeper Blue → Deep Blue
- deepfake
- deepfake audio → deepfake
- deepfake detection
- deepfake pornography → deepfake
- deepfake video → deepfake
- deepfakes → deepfake
- DeepLIFT
- DeepMind
- DeepMind Technologies → DeepMind
- DeepONet → neural operator
- DeepQA → IBM Watson
- DeepSeek
- DeepSeek moment
- DeepSeek shock → DeepSeek moment
- DeepSeek Sparse Attention → sparse attention
- DeepSeek-R1
- DeepSeek-R1-Zero → DeepSeek-R1
- DeepSHAP → SHAP
- DeepSpeed
- DeepSpeed library → DeepSpeed
- DeepSpeed Ulysses → context parallelism
- DeepSpeed-Chat → DeepSpeed
- DeepSpeed-MoE → DeepSpeed
- default box → anchor box
- default logic → non-monotonic reasoning
- default reasoning → non-monotonic reasoning
- defeasible reasoning → non-monotonic reasoning
- defence in depth → defense in depth
- Defense Advanced Research Projects Agency → DARPA
- defense in depth
- defensive accelerationism → differential technological development
- defer to trusted → trusted monitoring
- define-and-run → computational graph
- define-by-run → computational graph
- definite clause → Horn clause
- Deformable DETR → DETR
- defuzzification → fuzzy logic
- degenerate feedback loop → feedback loop
- degree of freedom → degrees of freedom
- degrees of freedom
- DeiT → vision transformer
- delayed generalization → grokking
- delayed reward → sparse reward
- delayed updates → gradient accumulation
- delete list → STRIPS
- deliberative alignment
- delta features → MFCC
- delta robot → industrial robot
- delta rule
- DeltaNet → linear attention
- delve
- delve into → delve
- delves → delve
- delving → delve
- demand forecasting → time series forecasting
- demographic parity
- Demonstrate–Search–Predict → DSPy
- demonstrations → few-shot prompting
- Dempster's rule of combination → Dempster–Shafer theory
- Dempster–Shafer theory
- Denavit–Hartenberg parameters → forward kinematics
- DENDRAL
- dendrogram → hierarchical clustering
- denial of wallet → sponge example
- Dennard scaling
- Daniel Dennett
- denoising autoencoder
- denoising diffusion implicit model
- denoising diffusion model → diffusion model
- denoising diffusion probabilistic model
- denoising (patching) → activation patching
- Dense → fully connected layer
- dense block → DenseNet
- dense captioning → image captioning
- dense correspondence → optical flow
- dense layer → fully connected layer
- dense passage retrieval → dense retrieval
- dense prediction → semantic segmentation
- dense representation → embedding
- dense retrieval
- dense reward → reward shaping
- densely connected convolutional network → DenseNet
- DenseNet
- DenseNet-121 → DenseNet
- density → probability density function
- density estimation
- density function → probability density function
- density-based clustering → DBSCAN
- density-based spatial clustering of applications with noise → DBSCAN
- dependency parsing → syntactic parsing
- deploy → model deployment
- deployer → provider and deployer
- deployment → model deployment
- depth → layer
- Depth Anything → depth estimation
- depth estimation
- depth map → depth estimation
- depth-first → depth-first search
- depth-first search
- depth-limited search → iterative deepening search
- depthwise convolution → depthwise separable convolution
- depthwise separable convolution
- DER → speaker diarization
- derivative
- derived intentionality → intentionality
- description logic
- description logics → description logic
- descriptive complexity → Kolmogorov complexity
- descriptor → feature extraction
- design stance → intentional stance
- deskill → deskilling
- deskilling
- detection head → backbone
- detection rate → recall
- DEtection TRansformer → DETR
- detection transformer → DETR
- detector → object detection
- deterministic policy gradient → deep deterministic policy gradient
- detokenization → tokenization
- detokenizer → tokenizer
- DETR
- dev interp → developmental interpretability
- dev set → validation set
- developer message → system prompt
- development set → validation set
- developmental interpretability
- Devin
- Devin AI → Devin
- devinterp → developmental interpretability
- dexterity → dexterous manipulation
- dexterous hand → dexterous manipulation
- dexterous manipulation
- DFS → depth-first search
- DFT → Fourier transform
- DGM → Darwin Gödel Machine
- DH parameters → forward kinematics
- diagnostic classifier → probing
- dialog system → conversational AI
- dialogue model → chat model
- dialogue system → conversational AI
- diarisation → speaker diarization
- diarization → speaker diarization
- diarization error rate → speaker diarization
- Dice coefficient → F1 score
- dictation → speech recognition
- dictionary learning
- Diederik P. Kingma → Diederik Kingma
- die-to-die interconnect → chiplet
- difference reduction → means–ends analysis
- difference-of-means probe → linear probe
- differentiable → derivative
- differentiable architecture search → neural architecture search
- differentiable neural computer → neural Turing machine
- differentiable programming → automatic differentiation
- differentiable simulator → physics simulator
- differential coefficient → derivative
- differential dynamic programming → optimal control
- differential entropy → entropy
- differential evolution → evolutionary computation
- differential privacy
- differential progress → differential technological development
- differential response modeling → uplift modeling
- differential technological development
- differential technology development → differential technological development
- differentially private → differential privacy
- differentially private SGD → DP-SGD
- differentially private stochastic gradient descent → DP-SGD
- differentiate → derivative
- differentiation → derivative
- diffing (interpretability) → model diffing
- diffusion → diffusion model
- diffusion distillation
- diffusion language model
- diffusion LLM → diffusion language model
- diffusion model
- diffusion policy
- diffusion probabilistic model → diffusion model
- diffusion sampler
- diffusion schedule → noise schedule
- diffusion transformer
- diffusion-based policy → diffusion policy
- digital assistant → virtual assistant
- digital colonialism → AI colonialism
- digital divide
- digital double → digital replica
- digital immortality → mind uploading
- digital inequality → digital divide
- digital mind
- digital minds → digital mind
- Digital Omnibus
- Digital Omnibus on AI → Digital Omnibus
- digital people → digital mind
- digital person → digital mind
- digital replica
- Digital Services Act
- digital shadow → digital twin
- digital simplification package → Digital Omnibus
- digital twin
- digital twins → digital twin
- digital welfare → AI welfare
- digital worker → robotic process automation
- Dijkstra algorithm → Dijkstra's algorithm
- Dijkstra's algorithm
- dilated causal convolution → WaveNet
- dilated convolution
- dilation rate → dilated convolution
- dilemma of control → Collingridge dilemma
- DiLoCo
- dimension → dimensionality
- dimension reduction → dimensionality reduction
- dimensionality
- dimensionality reduction
- dimensions → dimensionality
- DINO
- DINOv2 → DINO
- DINOv3 → DINO
- direct alignment → direct alignment algorithm
- direct alignment algorithm
- direct and indirect effects → causal mediation analysis
- direct collocation → trajectory optimization
- direct kinematics → forward kinematics
- direct liquid cooling → liquid cooling
- direct logit attribution
- direct preference optimization
- direct prompt injection → prompt injection
- directed acyclic graph → structural causal model
- directed graphical model → Bayesian network
- direction ablation → directional ablation
- directional ablation
- Directive (EU) 2024/2853 → Product Liability Directive
- direct-to-chip cooling → liquid cooling
- disaggregated inference → disaggregated serving
- disaggregated serving
- discontinuous takeoff → fast takeoff
- discount factor
- discount rate → discount factor
- discounted cumulative gain → normalized discounted cumulative gain
- discounted return → return
- discounting → discount factor
- discrete AdaBoost → AdaBoost
- discrete distribution → categorical distribution
- discrete Fourier transform → Fourier transform
- discriminant analysis → linear discriminant analysis
- discrimination by proxy → proxy discrimination
- discrimination information → Kullback–Leibler divergence
- discriminative classifier → discriminative model
- discriminative model
- disentangled representation
- disentangled representation learning → disentangled representation
- disentanglement → disentangled representation
- disinfo → disinformation
- disinformation
- disparate impact
- disparity → depth estimation
- distance metric learning → metric learning
- distant supervision → weak supervision
- DistilBERT → BERT
- Distill
- Distill journal → Distill
- distillation → knowledge distillation
- distillation attack → model extraction
- distilling context → context distillation
- distill.pub → Distill
- Distributed AI Research Institute
- distributed alignment search
- Distributed Low-Communication training → DiLoCo
- distributed representation → embedding
- Distributed Shampoo → Shampoo
- distributed training
- distributed word representation → word embedding
- DistributedDataParallel → data parallelism
- distribution → probability distribution
- distribution matching distillation → diffusion distillation
- distribution shift
- distributional hypothesis
- distributional semantics → distributional hypothesis
- distributional similarity → distributional hypothesis
- distribution-free method → nonparametric model
- DiT → diffusion transformer
- divergence → training instability
- divisive clustering → hierarchical clustering
- DLA → direct logit attribution
- DLG → gradient inversion attack
- dLLM → diffusion language model
- DLM → diffusion language model
- Dmitry Bahdanau → Dzmitry Bahdanau
- d_model → dimensionality
- D/N ratio → tokens per parameter
- DNA synthesis screening → biosecurity
- DNC → neural Turing machine
- DNN → deep learning
- Do Anything Now → DAN
- do-calculus
- DOCTOR → ELIZA
- document AI → document understanding
- document chunking → chunking
- document classification → text classification
- document embedding → text embedding
- document intelligence → document understanding
- document parsing → document understanding
- document understanding
- DocVQA → visual question answering
- DOF → degrees of freedom
- DoF → degrees of freedom
- Dojo → Tesla
- domain adaptation
- domain generalization → domain adaptation
- domain ontology → ontology
- domain randomisation → domain randomization
- domain randomization
- domain weights → data mixture
- domain-adaptive pretraining → continued pretraining
- domain-adversarial training → domain adaptation
- Donald O. Hebb → Donald Hebb
- Donald Olding Hebb → Donald Hebb
- do-not-train → opt-out
- Don't Create the Torment Nexus → Torment Nexus
- doomer
- doomerism → doomer
- doomers → doomer
- doom-pilled → -pilled
- do-operator → do-calculus
- dot product
- dot-product attention → scaled dot-product attention
- Doubao → ByteDance
- double descent
- Double DQN → deep Q-network
- double precision → floating-point format
- double Q-learning → Q-learning
- Doug Lenat → Douglas Lenat
- Douglas B. Lenat → Douglas Lenat
- Douglas R. Hofstadter → Douglas Hofstadter
- Douglas Richard Hofstadter → Douglas Hofstadter
- downloadable model → open-weight model
- Downpour SGD → parameter server
- downstream provider → provider and deployer
- downstream task → task (machine learning)
- do(x) → do-calculus
- DP training → DP-SGD
- DPG → deep deterministic policy gradient
- DPLL → Boolean satisfiability problem
- DPLL(T) → satisfiability modulo theories
- DPM-Solver → diffusion sampler
- DPO → direct preference optimization
- DPR → dense retrieval
- DP-SGD
- DQN → deep Q-network
- draft model → speculative decoding
- draft-and-verify decoding → speculative decoding
- DRAM
- DrawBench → Imagen
- DreamBooth
- DreamFusion → score distillation sampling
- Dreyfus critique → Dreyfus's critique of AI
- Hubert Dreyfus
- Dreyfus's critique of AI
- driver-assistance system → advanced driver-assistance system
- driverless car → autonomous vehicle
- driverless taxi → robotaxi
- DRL → deep reinforcement learning
- drone
- drone swarm → swarm robotics
- drop out → dropout
- DropConnect → dropout
- dropout
- dropout rate → dropout
- drop-upcycling → sparse upcycling
- DRY sampling → repetition penalty
- DS → data science
- d-separation → probabilistic graphical model
- DSPy
- DSPy optimizer → DSPy
- DSPy signature → DSPy
- dual encoder → two-tower model
- dual process theory → System 1 and System 2
- dual variable → Lagrange multiplier
- dual-encoder model → two-tower model
- dualism → mind–body problem
- dual-process theory → System 1 and System 2
- dual-use
- dual-use foundation model
- dual-use research of concern → dual-use
- dual-use technology → dual-use
- dummy coding → one-hot encoding
- dummy variable → one-hot encoding
- DURC → dual-use
- Durk Kingma → Diederik Kingma
- DUSt3R → 3D reconstruction
- DVC → data versioning
- Cynthia Dwork
- dying ReLU
- dying ReLU problem → dying ReLU
- Dyna
- Dyna architecture → Dyna
- dynamic Bayesian network → Bayesian network
- dynamic benchmark → contamination-resistant benchmark
- dynamic graph → computational graph
- dynamic programming
- dynamic programming equation → Bellman equation
- dynamic RAM → DRAM
- dynamic random-access memory → DRAM
- dynamic routing → capsule network
- dynamics model → world model
- dynamics randomization → domain randomization
- Dyna-Q → Dyna