VIEW THIS AS

Auto mode follows the Route Engine until you choose a viewpoint.

YOU ARE HERE

ROUTE CHECK

CONNECTED TO

WHAT NEXT

Use the canonical route for this room, or HELP if you are unsure.

Top 100 Vocabulary for Adults | AI & Machine Learning Professionals

Top 100 Vocabulary for Adults | AI & Machine Learning Professionals

AI and machine learning vocabulary is the language of systems that learn patterns from data and then act under uncertainty. Professionals need to distinguish training from inference, benchmark from real-world performance, model capability from product reliability, and fluent output from grounded evidence.

This profession-specific flagship belongs to the eduKate Adult Vocabulary for Professionals system. It complements Data & Analytics Professionals, Software & Technology Professionals and Cybersecurity Professionals.

The Four Banks

Models & Learning: artificial intelligence, machine learning, model, algorithm, parameter, feature, label, training, inference, supervised learning, unsupervised learning, reinforcement learning, classification, regression, clustering, neural network, transformer, embedding, token, context window, fine-tuning, pretraining, loss, optimisation, gradient.

Data & Evaluation: dataset, training set, validation set, test set, ground truth, annotation, sampling, imbalance, leakage, benchmark, metric, accuracy, precision, recall, F1 score, calibration, robustness, generalisation, overfitting, underfitting, hallucination, uncertainty, confidence, evaluation, red teaming.

Generative AI & Agents: prompt, completion, system instruction, tool call, retrieval, RAG, vector database, semantic search, agent, workflow, planning, memory, grounding, citation, guardrail, structured output, multimodal, temperature, sampling, reasoning, chain, orchestration, latency, token cost, context.

Deployment, Safety & Operations: deployment, endpoint, API, model serving, throughput, latency, scaling, monitoring, drift, observability, logging, feedback loop, human-in-the-loop, fallback, rollback, versioning, governance, privacy, security, bias, fairness, explainability, alignment, safety, reliability.

Top 100 AI & Machine Learning Vocabulary: Working Meanings

#WordProfessional meaning
1Artificial intelligenceComputer systems performing tasks associated with human cognitive capabilities.
2Machine learningMethods allowing systems to learn patterns from data rather than rely only on explicit rules.
3ModelA learned mathematical representation used to make predictions or generate outputs.
4AlgorithmA defined computational procedure for solving a problem.
5ParameterA learned numerical value controlling model behaviour.
6FeatureAn input variable used by a model.
7LabelA target value associated with a training example.
8TrainingThe process of adjusting model parameters using data and an objective.
9InferenceUsing a trained model to produce predictions or outputs.
10Supervised learningLearning from examples paired with target labels.
11Unsupervised learningLearning patterns from data without explicit target labels.
12Reinforcement learningLearning behaviour through interaction and reward signals.
13ClassificationPredicting a category.
14RegressionPredicting a numerical value.
15ClusteringGrouping similar observations without predefined labels.
16Neural networkA layered computational model composed of interconnected learned units.
17TransformerA neural-network architecture based heavily on attention mechanisms.
18EmbeddingA numerical vector representing semantic or structural information.
19TokenA unit of text or another modality processed by a model.
20Context windowThe amount of input and generated content a model can consider at once.
21Fine-tuningAdditional training adapting a pretrained model to particular tasks or behaviours.
22PretrainingLarge-scale initial training before task-specific adaptation.
23LossA numerical objective representing model error during training.
24OptimisationThe process of adjusting parameters to reduce loss or improve an objective.
25GradientInformation describing how model parameters should change to alter the objective.
26DatasetA structured collection of examples used for training or evaluation.
27Training setData used to fit model parameters.
28Validation setData used to tune choices during model development.
29Test setHeld-out data used for final evaluation.
30Ground truthThe reference answer or outcome treated as correct for evaluation.
31AnnotationHuman- or machine-assigned labels or metadata added to data.
32SamplingSelecting data points or model outputs according to a procedure.
33ImbalanceUnequal representation of categories or outcomes in a dataset.
34LeakageUnintended use of information unavailable at real prediction time.
35BenchmarkA standard test used to compare model performance.
36MetricA quantitative measure of model performance.
37AccuracyThe proportion of predictions that are correct.
38PrecisionThe proportion of positive predictions that are truly positive.
39RecallThe proportion of true positives that are successfully identified.
40F1 scoreThe harmonic mean of precision and recall.
41CalibrationThe degree to which predicted confidence matches observed outcomes.
42RobustnessAbility to maintain performance under variation or perturbation.
43GeneralisationAbility to perform well on previously unseen data.
44OverfittingLearning training-specific patterns that do not generalise well.
45UnderfittingFailing to learn enough structure to perform well even on training data.
46HallucinationA generated output containing unsupported or fabricated information.
47UncertaintyIncomplete knowledge about model correctness or future outcomes.
48ConfidenceA model or system estimate of certainty, which may or may not be calibrated.
49EvaluationStructured measurement of model behaviour against defined criteria.
50Red teamingAdversarial testing intended to uncover failures, misuse or vulnerabilities.
51PromptInput instructing or conditioning a generative model.
52CompletionText or other content generated in response to input.
53System instructionA high-priority instruction defining model behaviour and constraints.
54Tool callA structured request from a model to an external function or service.
55RetrievalFetching relevant external information for use during generation.
56RAGRetrieval-augmented generation: combining external retrieval with model generation.
57Vector databaseA database designed to store and search numerical embeddings.
58Semantic searchRetrieval based on meaning rather than exact keyword matching alone.
59AgentA model-driven system that can plan, choose actions and use tools toward a goal.
60WorkflowA defined sequence of AI and non-AI steps producing an outcome.
61PlanningSelection and ordering of actions toward a goal.
62MemoryStored information used to preserve relevant context across steps or interactions.
63GroundingConnecting generated outputs to reliable external evidence or state.
64CitationA reference linking a claim to supporting source material.
65GuardrailA control constraining model or application behaviour.
66Structured outputModel output conforming to a predefined schema or format.
67MultimodalCapable of processing or generating more than one type of media.
68TemperatureA sampling parameter influencing output randomness.
69SamplingThe process of selecting the next output from a probability distribution.
70ReasoningInternal or explicit computation used to reach a conclusion.
71ChainA sequence of model, tool or transformation steps.
72OrchestrationCoordination of models, tools and workflows.
73LatencyThe delay between request and usable output.
74Token costThe computational or monetary cost associated with processed tokens.
75ContextInformation available to the model for interpreting the current task.
76DeploymentPutting a model or AI application into operational use.
77EndpointA network-accessible interface for invoking a model or service.
78APIAn application programming interface allowing software systems to interact.
79Model servingInfrastructure used to deliver model inference to applications.
80ThroughputThe number of requests or tokens processed over time.
81ScalingIncreasing system capacity to handle more workload.
82MonitoringOngoing observation of model and application behaviour.
83DriftChange in data, behaviour or relationships that may reduce model performance.
84ObservabilityThe ability to understand internal system state from logs, metrics and traces.
85LoggingRecording system events and outputs for analysis.
86Feedback loopA process using observed outcomes to improve future behaviour.
87Human-in-the-loopA system design requiring human review or decision at specified points.
88FallbackAn alternative action used when the preferred model or workflow fails.
89RollbackReverting to an earlier model or system version after problems.
90VersioningTracking distinct model, prompt, data or application versions.
91GovernanceThe structure of ownership, oversight and accountability for AI systems.
92PrivacyAppropriate collection, use and protection of personal information.
93SecurityProtection of AI systems, data and users from malicious or accidental harm.
94BiasSystematic patterns producing distorted or unfair outcomes.
95FairnessAssessment of whether outcomes treat relevant groups appropriately under stated criteria.
96ExplainabilityThe degree to which model behaviour or decisions can be meaningfully understood.
97AlignmentThe degree to which model behaviour matches intended objectives and constraints.
98SafetyReduction of unacceptable risk from AI system behaviour or use.
99ReliabilityAbility of an AI system to perform as intended consistently.
100Model riskThe possibility of harm or loss arising from incorrect, unstable or misused model behaviour.

Benchmark Performance Is Not Product Reliability

A model can score well on a benchmark and still fail in production because users ask different questions, retrieval is weak, data drifts, latency is unacceptable or the application lacks fallbacks. The unit of reliability is the whole system, not just the model.

Scenario: Fluent but Unsupported Answer

Check whether the model had access to the required evidence, whether retrieval surfaced it, whether citations actually support the claim and whether the workflow should have abstained. Fluency is a presentation property; grounding is an evidence property.

Seven-Day AI & Machine Learning Vocabulary Plan

DayPractice
1Separate model, training, inference and evaluation.
2Map datasets, labels, leakage and generalisation.
3Compare accuracy, precision, recall and calibration.
4Build a retrieval-grounded generative workflow.
5Audit one AI application for fallback, monitoring and safety.
6Recall 75+ AI terms by function.
7Write a one-page AI system review separating model capability from operational reliability.

Continue the Digital Profession Wing

Conclusion

AI and machine learning vocabulary helps professionals make a fast-moving field more exact. It separates model from application, training from inference, confidence from correctness, and impressive demos from systems that can be trusted under real conditions.

Discover more from eduKate Singapore

Subscribe now to keep reading and get access to the full archive.

Continue reading