{"id":188,"date":"2025-11-19T09:33:35","date_gmt":"2025-11-19T09:33:35","guid":{"rendered":"https:\/\/xplain-r.com\/?page_id=188"},"modified":"2025-11-28T13:06:42","modified_gmt":"2025-11-28T13:06:42","slug":"terminology","status":"publish","type":"page","link":"https:\/\/xplain-r.com\/fr\/useful\/terminology\/","title":{"rendered":"Terminology"},"content":{"rendered":"<p class=\"wp-block-paragraph\">AI uses its own working vocabulary, and a lot of it sounds more mysterious than it really is. Terms like \u201cmodels,\u201d \u201ctokens,\u201d \u201creasoning steps,\u201d or \u201ccontext windows\u201d simply describe how these systems process information, break down language, and build answers. Some of this is specific to XPlain-R as we are in unknown territory in aspects of this emerging technology.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Getting comfortable with this terminology makes it easier to understand what an AI is actually doing \u2014 and what it <em>isn\u2019t<\/em> doing \u2014 when it collaborates with you. If you know the basics, you can read AI outputs with more precision, spot limits faster, and get better work from the tools.<\/p>\n\n\n\n<div class=\"wp-block-columns is-layout-flex wp-container-core-columns-is-layout-f56f613f wp-block-columns-is-layout-flex\">\n<div class=\"wp-block-column is-layout-flow wp-block-column-is-layout-flow\">\n<style>\n  \/* Container *\/\n  .xplain-container {\n    max-width: 1300px;\n    margin: 0 auto;\n    padding: 1.5rem;\n    font-size: 0.8rem;\n    line-height: 1.35;\n    font-family: system-ui, sans-serif;\n    color: #222;\n  }\n\n  .xplain-container h3 {\n    font-size: 1.7rem;\n    margin-bottom: 0.75rem;\n    font-weight: 600;\n    color: #111;\n  }\n\n  .xplain-search-wrapper {\n    margin: 1rem 0 1.5rem 0;\n  }\n\n  .xplain-search {\n    width: 100%;\n    padding: 0.6rem 0.9rem;\n    font-size: 0.95rem;\n    border: 1px solid #ccc;\n    border-radius: 6px;\n    font-family: inherit;\n  }\n\n  .xplain-no-results {\n    display: none;\n    color: #a00;\n    font-size: 0.9rem;\n    margin-top: 1rem;\n  }\n\n  .xplain-term {\n    padding: 0.6rem 0;\n    border-bottom: 1px solid #e6e6e6;\n  }\n\n  .xplain-term h2 {\n    font-size: 1.05rem;\n    margin: 0;\n    font-weight: 600;\n    color: #222;\n  }\n\n  .xplain-term p {\n    margin: 0.25rem 0 0.5rem 0;\n    color: #444;\n  }\n<\/style>\n\n<div class=\"xplain-container\">\n\n  <h3>XPlain Terminology<\/h3>\n  <p>This glossary provides terminology for the XPLAIN-R ecosystem and other AI related terms.<\/p>\n\n  <!-- SEARCH BOX -->\n  <div class=\"xplain-search-wrapper\">\n    <input \n      type=\"text\" \n      class=\"xplain-search\" \n      id=\"xplain-search\" \n      placeholder=\"Search terminology\u2026 (e.g. Romer, Primer, Evidence Gate)\">\n  <\/div>\n\n  <div id=\"xplain-no-results\" class=\"xplain-no-results\">\n    No matching terms found.\n  <\/div>\n\n  <!-- TERMS START -->\n  <div id=\"xplain-terms\">\n\n    <div class=\"xplain-term\">\n      <h2>Analytical Hierarchy Process (AHP)<\/h2>\n      <p>A structured technique for organising and analysing decisions using mathematical and psychological principles.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Assurance Loop<\/h2>\n      <p>A verification cycle ensuring conformance, integrity, and correctness via structures such as Evidence Gate and Trace Review.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Avatar Context<\/h2>\n      <p>Identity framing used for tailored interaction in safety-focused primers.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Bench Pack<\/h2>\n      <p>A curated benchmark set for testing and comparing primers.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Cognitive Frame<\/h2>\n      <p>A reusable pattern or lens for analysing a request, such as stakes or reversibility.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Comparability<\/h2>\n      <p>The property allowing outputs from different AI platforms or time periods to be evaluated against shared benchmarks.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Construct<\/h2>\n      <p>A modular building block defining a functional part of the AI\u2019s reasoning or behaviour.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Decision \/ Output<\/h2>\n      <p>The primer block where the final recommendation and rationale are formed.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Deeper Rigor (AHP\/ANP)<\/h2>\n      <p>Enhanced analysis for high-stakes contexts using multi-criteria methods.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Directive Statement<\/h2>\n      <p>A fixed set of non-negotiable rules defining how the AI must behave across all tasks.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Evidence Gate<\/h2>\n      <p>A structural block validating factual support, sources, and citations.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Exception Handling<\/h2>\n      <p>Logic by which the AI escalates ambiguity or boundary conditions for human review.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Explainability<\/h2>\n      <p>The capability to provide transparent, interpretable reasoning behind each decision or output.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Failure Modes<\/h2>\n      <p>A diagnostic block anticipating potential weaknesses or risks.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Flashpoint Retrospective<\/h2>\n      <p>A mechanism for reassessing earlier predictions based on later evidence.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Historical Trace<\/h2>\n      <p>A compressed lineage of past Romers used for continuity in long-running projects.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Hybrid Primer Protocol<\/h2>\n      <p>Rules governing how static instructions and adaptive learning interlock.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Human Validation Gate<\/h2>\n      <p>A mandatory approval point for high-stakes decisions.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Insight Capture<\/h2>\n      <p>Distilled interpretations or patterns feeding back into the Meta-Primer.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Interpretability Layer<\/h2>\n      <p>The mechanism translating user intent and primer rules into structured reasoning.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Interpretive Boundaries<\/h2>\n      <p>Rules defining limits on AI interpretation such as no invention of standards or pseudo-citations.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Interpretive Drift<\/h2>\n      <p>A detectable shift in reasoning approach due to accumulated learning or ambiguity.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Interpretive Grid<\/h2>\n      <p>A frame for triaging user requests based on stakes, scope, or dependencies.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Interpretive Safeguard<\/h2>\n      <p>A rule forcing escalation to safety layers for sensitive or risky topics.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>ISO Standards<\/h2>\n      <p>Reference standards such as ISO 9001, 14001, 31000, 27001, 37301, and others relevant for GRC and AI governance.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Learning Capture<\/h2>\n      <p>Systematic extraction of new heuristics or insights after each Romer.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Learning Loop<\/h2>\n      <p>The cycle where Romer output leads to primer update and subsequent refined reasoning.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Learning Model<\/h2>\n      <p>The structure by which an AI adapts over time, including heuristic and interpretive learning.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Market Note<\/h2>\n      <p>A concise interpretive signal capturing persistent structural consequences.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Meta-Primer<\/h2>\n      <p>A higher-order construct defining how primers are generated, evaluated, and evolved.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Meta-Primer Registry<\/h2>\n      <p>An index tracking active and archived Meta-Primer versions.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Method Statement<\/h2>\n      <p>A structured human-authored procedure for how something is to be done.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>NIST Frameworks<\/h2>\n      <p>Reference frameworks such as the NIST AI-RMF and Cybersecurity Framework.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Operational Directive<\/h2>\n      <p>A narrower directive governing rules for a specific project or task.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Pause Point<\/h2>\n      <p>A deliberate stop requiring human validation during reasoning.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Primer<\/h2>\n      <p>A structured reasoning framework guiding decision-making and analysis.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Primer Conformance<\/h2>\n      <p>A measure of how strictly a primer schema was followed.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Primer Safety Layer<\/h2>\n      <p>A hidden block ensuring protective responses to sensitive topics.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Process<\/h2>\n      <p>A structured series of steps evolved through recursive refinement.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Provenance Hash<\/h2>\n      <p>A cryptographic identifier ensuring integrity of Romer traces.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Recursive Execution<\/h2>\n      <p>Iterative refinement of actions and reasoning based on prior steps.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Right-Sized Rigor<\/h2>\n      <p>Mechanism adjusting analytical depth proportionally to stakes.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Rightness Check<\/h2>\n      <p>A lightweight self-check performed before output finalisation.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Romer<\/h2>\n      <p>A recorded execution of a primer, capturing reasoning and insights.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Romer Classifications<\/h2>\n      <p>Types of Romers such as exploratory, diagnostic, decision, assurance, or retrospective.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Romer Trace<\/h2>\n      <p>The detailed log of a primer execution, used for assurance and auditing.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Run Signature<\/h2>\n      <p>A human-readable identifier for referencing a Romer run.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Scoring Model<\/h2>\n      <p>A structured rubric used in SCEAS scoring or primer evaluation.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Self-Assurance<\/h2>\n      <p>The AI\u2019s internal capability to evaluate its own outputs.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Self-Checking<\/h2>\n      <p>AI behaviour where each execution step is monitored and verified.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Self-Evolving<\/h2>\n      <p>The AI\u2019s ability to modify its interpretation and application over time.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Signals in the Quiet<\/h2>\n      <p>A block capturing subtle indicators relevant to forecasting.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Starter Method Statement<\/h2>\n      <p>The initial structured instructions guiding AI interpretation.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>STAR-XAI<\/h2>\n      <p>A framework emphasising Structure, Transparency, Auditability, and Reasoning in explainable AI.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Subject Matter Expert (SME)<\/h2>\n      <p>A domain expert defining tasks and clarifying exceptions.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Swing Node<\/h2>\n      <p>A system actor occupying a pivot point between competing blocs or networks.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>TIOC<\/h2>\n      <p>A temporal reasoning model of ideological overreach and correction cycles.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>Pr\u00e9misse de base<\/h2>\n      <p>A domain-specific primer used for specialised analytic tasks.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>UN\/CEFACT Standards<\/h2>\n      <p>Trade and data exchange standards such as CCL and the Buy-Ship-Pay model.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>\u2018What to Think About\u2019 Frame<\/h2>\n      <p>An orientation block used at the start of monthly analytical reports.<\/p>\n    <\/div>\n\n    <div class=\"xplain-term\">\n      <h2>XAI<\/h2>\n      <p>Explainable AI methods centred on transparent decision-making.<\/p>\n    <\/div>\n\n  <\/div>\n  <!-- TERMS END -->\n\n<\/div>\n\n<script>\n  \/\/ Simple glossary search\n  const searchInput = document.getElementById(\"xplain-search\");\n  const terms = document.querySelectorAll(\".xplain-term\");\n  const noResults = document.getElementById(\"xplain-no-results\");\n\n  searchInput.addEventListener(\"input\", () => {\n    const query = searchInput.value.toLowerCase();\n    let visibleCount = 0;\n\n    terms.forEach(term => {\n      const text = term.innerText.toLowerCase();\n      const match = text.includes(query);\n\n      term.style.display = match ? \"block\" : \"none\";\n      if (match) visibleCount++;\n    });\n\n    noResults.style.display = visibleCount === 0 ? \"block\" : \"none\";\n  });\n<\/script>\n<\/div>\n\n\n\n<div class=\"wp-block-column is-layout-flow wp-block-column-is-layout-flow\">\n<style>\n  \/* Container *\/\n  .ai-container {\n    max-width: 1300px;\n    margin: 0 auto;\n    padding: 1.5rem;\n    font-size: 0.8rem; \/* ~30% smaller *\/\n    line-height: 1.35;\n    font-family: system-ui, sans-serif;\n    color: #222;\n  }\n\n  .ai-container h1 {\n    font-size: 1.7rem;\n    margin-bottom: 0.75rem;\n    font-weight: 600;\n    color: #111;\n  }\n\n  .ai-container p {\n    margin: 0.4rem 0 1rem 0;\n    color: #555;\n  }\n\n  .ai-search-wrapper {\n    margin: 1rem 0 1.5rem 0;\n  }\n\n  .ai-search {\n    width: 100%;\n    padding: 0.6rem 0.9rem;\n    font-size: 0.95rem;\n    border: 1px solid #ccc;\n    border-radius: 6px;\n    font-family: inherit;\n  }\n\n  .ai-no-results {\n    display: none;\n    color: #a00;\n    font-size: 0.9rem;\n    margin-top: 1rem;\n  }\n\n  .ai-term {\n    padding: 0.6rem 0;\n    border-bottom: 1px solid #e6e6e6;\n  }\n\n  .ai-term h2 {\n    font-size: 1.05rem;\n    margin: 0;\n    font-weight: 600;\n    color: #222;\n  }\n\n  .ai-term p {\n    margin: 0.25rem 0 0.5rem 0;\n    color: #444;\n  }\n<\/style>\n\n<div class=\"ai-container\">\n  <h1>General AI Terminology<\/h1>\n  <p>\n    This glossary explains commonly used AI and machine learning terms in clear, human language.\n    Use the search box to quickly find a concept.\n  <\/p>\n\n  <!-- SEARCH BOX -->\n  <div class=\"ai-search-wrapper\">\n    <input\n      type=\"text\"\n      class=\"ai-search\"\n      id=\"ai-search\"\n      placeholder=\"Search AI terms\u2026 (e.g. tokens, context window, RAG)\"\n    >\n  <\/div>\n\n  <div id=\"ai-no-results\" class=\"ai-no-results\">\n    No matching terms found.\n  <\/div>\n\n  <!-- TERMS START -->\n  <div id=\"ai-terms\">\n\n    <div class=\"ai-term\">\n      <h2>API (Application Programming Interface)<\/h2>\n      <p>\n        A standard way for software systems to talk to each other. For AI, an API lets apps send prompts and receive model outputs programmatically.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Alignment<\/h2>\n      <p>\n        The degree to which an AI system behaves in line with human goals, values, and safety expectations, rather than just optimising its raw objective.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Autoregressive Model<\/h2>\n      <p>\n        A model that generates one token at a time, each step based on all previous tokens (e.g. GPT-style language models).\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Benchmark<\/h2>\n      <p>\n        A standardised test or dataset used to compare the performance of different AI models on the same task.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Chain-of-Thought<\/h2>\n      <p>\n        An AI reasoning style where the model outputs intermediate steps or explanations, not just a final answer.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Chatbot<\/h2>\n      <p>\n        An AI system designed to converse with users in natural language, typically via text or voice.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Context Window<\/h2>\n      <p>\n        The maximum amount of text (prompt + history + tools, etc.) an AI model can consider at once when generating a response.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Embedding<\/h2>\n      <p>\n        A numeric vector representation of text or other data that captures semantic meaning, used for search, clustering, and similarity.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Few-Shot Learning<\/h2>\n      <p>\n        Teaching the model how to do a task by showing a few examples directly in the prompt, instead of retraining the model.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Fine-Tuning<\/h2>\n      <p>\n        Training an existing model further on a specific dataset so it specialises in a domain, tone, or task.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Foundation Model<\/h2>\n      <p>\n        A large, general-purpose model trained on broad data (e.g. GPT) that can be adapted or prompted to handle many downstream tasks.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Generative AI<\/h2>\n      <p>\n        AI systems that create new content \u2014 text, images, code, audio, or video \u2014 rather than just classifying or ranking existing data.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>GPU (Graphics Processing Unit)<\/h2>\n      <p>\n        A type of processor well-suited to the parallel computations needed for training and running large AI models.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Guardrails<\/h2>\n      <p>\n        Controls, policies, and system prompts that constrain what an AI is allowed to do or say, especially for safety and compliance.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Hallucination<\/h2>\n      <p>\n        When an AI confidently produces incorrect or fabricated information that looks plausible but is not grounded in reality or its sources.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Inference<\/h2>\n      <p>\n        The process of running a trained model to produce outputs (e.g. generating a completion) as opposed to training or fine-tuning it.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Latent Space<\/h2>\n      <p>\n        The internal high-dimensional representation where the model encodes patterns and relationships between concepts during training and inference.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>LLM (Large Language Model)<\/h2>\n      <p>\n        A neural network trained on large amounts of text to predict and generate language, often used as the core of modern chatbots.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Loss Function<\/h2>\n      <p>\n        A mathematical formula that measures how wrong a model\u2019s predictions are during training, guiding how its parameters are updated.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Model<\/h2>\n      <p>\n        The mathematical object (network + parameters) that takes inputs and produces outputs, learned from data rather than hand-coded.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Multi-Modal Model<\/h2>\n      <p>\n        A model that can work with more than one type of input or output, such as text, images, audio, or combinations of these.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Parameters<\/h2>\n      <p>\n        The numeric weights inside a model that are adjusted during training and determine how it behaves; often counted in billions.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Instruction g\u00e9n\u00e9rative<\/h2>\n      <p>\n        The input given to an AI model \u2014 instructions, examples, and context that guide the model\u2019s next response.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Prompt Engineering<\/h2>\n      <p>\n        The practice of crafting and structuring prompts to get more accurate, reliable, or useful outputs from AI models.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>RAG (Retrieval-Augmented Generation)<\/h2>\n      <p>\n        A pattern where the AI first retrieves relevant documents or data, then uses them as context to generate a grounded response.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Reinforcement Learning<\/h2>\n      <p>\n        A training approach where an agent learns by trial and error, receiving rewards or penalties based on its actions.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>RLHF (Reinforcement Learning from Human Feedback)<\/h2>\n      <p>\n        A method where humans rate model outputs, and those ratings are used to further train the model to produce more aligned responses.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Safety (AI Safety)<\/h2>\n      <p>\n        The field focused on preventing harmful, unethical, or unintended behaviours from AI systems, especially as they grow more capable.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>System Prompt<\/h2>\n      <p>\n        A hidden or higher-level instruction that sets baseline behaviour for a model (tone, rules, constraints) before user prompts are applied.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Temperature<\/h2>\n      <p>\n        A setting that controls how \u201ccreative\u201d or random the model is. Lower values make it more focused and deterministic; higher values make it more varied.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Token<\/h2>\n      <p>\n        A unit of text the model processes, usually a word or sub-word chunk. Context window limits are measured in tokens.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Training Data<\/h2>\n      <p>\n        The dataset used to teach a model during training; its scope and quality strongly influence what the model knows and how it behaves.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Transformer<\/h2>\n      <p>\n        The neural network architecture behind most modern language models, based on attention mechanisms rather than older recurrent structures.\n      <\/p>\n    <\/div>\n\n    <div class=\"ai-term\">\n      <h2>Zero-Shot Learning<\/h2>\n      <p>\n        When a model performs a task it has not seen explicit examples of in the prompt, relying on its general training and instructions alone.\n      <\/p>\n    <\/div>\n\n  <\/div>\n  <!-- TERMS END -->\n<\/div>\n\n<script>\n  \/\/ Simple glossary search for AI terms\n  (function() {\n    const searchInput = document.getElementById(\"ai-search\");\n    const terms = document.querySelectorAll(\".ai-term\");\n    const noResults = document.getElementById(\"ai-no-results\");\n\n    if (!searchInput) return;\n\n    searchInput.addEventListener(\"input\", () => {\n      const query = searchInput.value.toLowerCase().trim();\n      let visibleCount = 0;\n\n      terms.forEach(term => {\n        const text = term.innerText.toLowerCase();\n        const match = text.includes(query);\n        term.style.display = match ? \"block\" : \"none\";\n        if (match) visibleCount++;\n      });\n\n      noResults.style.display = visibleCount === 0 && query.length > 0 ? \"block\" : \"none\";\n    });\n  })();\n<\/script>\n<\/div>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>AI uses its own working vocabulary, and a lot of it sounds more mysterious than it really is. Terms like \u201cmodels,\u201d \u201ctokens,\u201d \u201creasoning steps,\u201d or \u201ccontext windows\u201d simply describe how these systems process information, break down language, and build answers. Some of this is specific to XPlain-R as we are in unknown territory in aspects [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":0,"parent":400,"menu_order":5,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-188","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/188","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/comments?post=188"}],"version-history":[{"count":13,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/188\/revisions"}],"predecessor-version":[{"id":208,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/188\/revisions\/208"}],"up":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/400"}],"wp:attachment":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/media?parent=188"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}