A (52 terms)
A* Algorithm
Optimal pathfinding algorithm.
Advanced
A/B Testing
Controlled experiment comparing variants by random assignment to estimate causal effects of changes.
Intermediate
Absolute Positional Encoding
Encodes token position explicitly, often via sinusoids.
Intermediate
Accuracy
Fraction of correct predictions; can be misleading on imbalanced datasets.
Intermediate
Acoustic Model
Maps audio signals to linguistic units.
Intermediate
Action Space
Set of all actions available to the agent.
Intermediate
Activation Function
Nonlinear functions enabling networks to approximate complex mappings; ReLU variants dominate modern DL.
Intermediate
Active Experimentation
AI selecting next experiments.
Advanced
Active Inference
Acting to minimize surprise or free energy.
Frontier
Active Learning
Selecting the most informative samples to label (e.g., uncertainty sampling) to reduce labeling cost.
Intermediate
Actor-Critic
Combines value estimation (critic) with policy learning (actor).
Intermediate
Actuator
Hardware components that execute physical actions.
Advanced
Adam
Popular optimizer combining momentum and per-parameter adaptive step sizes via first/second moment estimates.
Intermediate
Adaptive Optimization
Methods like Adam adjusting learning rates dynamically.
Intermediate
Adversarial Example
Inputs crafted to cause model errors or unsafe behavior, often imperceptible in vision or subtle in text.
Intermediate
Adversarial Market
Market reacting strategically to AI.
Advanced
Affordance
Perceived actions an environment allows.
Frontier
Agent
A system that perceives state, selects actions, and pursues goals—often combining LLM reasoning with tools and memory.
Intermediate
Agent Loop
Continuous cycle of observation, reasoning, action, and feedback.
Intermediate
AI Adoption Curve
Organizational uptake of AI technologies.
Intermediate
AI Boxing
Isolating AI systems.
Advanced
AI Center of Excellence
Centralized AI expertise group.
Intermediate
AI Hallucination
Fabrication of cases or statutes by LLMs.
Intermediate
AI Treaty
International agreements on AI.
Intermediate
Algorithmic Accountability
Ensuring decisions can be explained and traced.
Intermediate
Algorithmic Bias
Unequal performance across demographic groups.
Intermediate
Algorithmic Collusion
AI tacitly coordinating prices.
Advanced
Algorithmic Trading
AI-driven buying/selling of financial assets.
Intermediate
Alignment
Ensuring model behavior matches human goals, norms, and constraints, including reducing harmful or deceptive outputs.
Intermediate
Alignment Problem
Ensuring AI systems pursue intended human goals.
Advanced
Alignment Research
Research ensuring AI remains safe.
Intermediate
Alignment Tax
Tradeoff between safety and performance.
Advanced
Alpha
Returns above benchmark.
Intermediate
AlphaFold
Deep learning system for protein structure prediction.
Advanced
ARIMA
Classical statistical time-series model.
Intermediate
Artificial General Intelligence
AI capable of performing most intellectual tasks humans can.
Frontier
Artificial Intelligence
The field of building systems that perform tasks associated with human intelligence—perception, reasoning, language, ...
Intermediate
Attention
Mechanism that computes context-aware mixtures of representations; scales well and captures long-range dependencies.
Intermediate
Attention Head
A single attention mechanism within multi-head attention.
Intermediate
Attorney-Client Privilege
Protection of private legal communications.
Intermediate
AUC
Scalar summary of ROC; measures ranking ability, not calibration.
Intermediate
Auction Mechanism
Rules governing auctions.
Advanced
Audit
Systematic review of model/data processes to ensure performance, fairness, security, and policy compliance.
Intermediate
Audit Trail
Logged record of model inputs, outputs, and decisions.
Intermediate
Auditability
Ability to inspect and verify AI decisions.
Intermediate
Autoencoder
Model that compresses input into latent space and reconstructs it.
Advanced
Automated Hypothesis Generation
AI proposing scientific hypotheses.
Advanced
Automation Bias
Tendency to trust automated suggestions even when incorrect; mitigated by UI design, training, and checks.
Intermediate
Autonomous Agent
System that independently pursues goals over time.
Advanced
Autoregressive Model
Generates sequences one token at a time, conditioning on past tokens.
Intermediate
Autoscaling
Dynamic resource allocation.
Intermediate
Average Treatment Effect
Expected causal effect of a treatment.
Advanced