Every AI word,
in plain English.

A

• A2AAgent2Agent protocol Agent2Agent: an open protocol that lets AI agents built by different companies discover…
• abliterationabliterated model a white-box model-editing technique that identifies an internal activation direction…
• accelerationisme/acc in tech culture, the view that AI and technology should be developed as fast as possible…
• activation monitoringactivation-based monitoring a technique for overseeing an AI model by examining patterns in its internal neural…
• activation patchingcausal tracing a mechanistic-interpretability technique that replaces selected internal activations from…
• activation steeringactivation-space steering a method for changing an AI model's behavior during inference by modifying selected…
• adversarial distillationillicit distillation the systematic, unauthorized use of a capable AI model's outputs or reasoning traces to…
• agent bill of materialsAgBOM a structured inventory of the components and capabilities that make up an AI agent, such as…
• agent containmentAI agent containment the use of enforced technical boundaries to limit what an AI agent can read, change…
• Agent Control StandardACS an open protocol for placing portable policy checks at points in an AI agent's runtime…
• agent experienceAX the quality of an AI agent's interaction with a website, application, API, or workflow, and…
• agent fleetAI agent fleet a group of AI agents deployed and managed as one operational system, often to carry out…
• agent gatewayAI agent gateway an infrastructure layer placed between AI agents and the models, tools, APIs, data, or…
• agent hijackingAI agent hijacking an attack that redirects an AI agent's goals, reasoning, or tool actions toward an…
• agent identityAI agent identity the identifiers, credentials, provenance, permissions, and delegated authority used to…
• agent loop the repeating cycle at the heart of an agent: the model decides on an action, a tool…
• agent sandboxAI agent sandbox an isolated, policy-controlled computing environment in which an AI agent can run commands…
• agent skillskill a packaged set of instructions, scripts, and resources that an AI agent can load when…
• agent sprawlAI agent sprawl the uncontrolled proliferation of AI agents across an organization faster than the…
• agent swarmswarm agents a large group of AI agents working on parts of a task at the same time, coordinated by a…
• agent washingagent-washing the practice of marketing or relabeling a chatbot, assistant, scripted workflow, or…
• agentic able to act independently toward a goal, taking multiple steps and making decisions with…
• agentic advertisingagentic buying an advertising model in which AI agents plan, negotiate, buy, sell, optimize, or measure…
• agentic browserAI browser a web browser with a built-in AI agent that can read pages and act for you, such as filling…
• agentic coding software development in which an AI agent plans changes, edits many files, runs tests and…
• agentic commerceagent commerce shopping in which AI agents search, compare, and buy products or services on a person's…
• agentic financeagent-based finance the use of autonomous or semi-autonomous AI agents to analyze financial situations, make…
• agentic operating systemagentic OS a software or infrastructure layer that manages the execution, memory, tools, permissions…
• agentic paymentsAI agent payments payments that an AI agent initiates or completes on behalf of a person, organization, or…
• agentic SOCagentic security operations center a security-operations-center model in which AI agents coordinate multi-step threat…
• agentic workflow a process in which one or more AI agents carry out a series of steps, such as research…
• AI agentagent an AI system that can pursue a goal over many steps on its own, deciding what to do next…
• AI agent insuranceAI agent liability insurance commercial insurance designed to cover specified losses or liabilities caused by autonomous…
• AI assistant an AI product that helps a person with everyday tasks such as answering questions, writing…
• AI bubbleartificial intelligence bubble the debated claim that investment, spending, or market valuations associated with…
• AI coding assistant a tool that helps programmers by suggesting code as they type, explaining code, writing…
• AI companion an AI character or chatbot designed for ongoing personal conversation, friendship, or…
• AI consciousnessmachine consciousness the question of whether an AI system could have subjective experience, meaning that there…
• AI constitutionmodel constitution a written document that sets out the values, priorities, and principles an AI model is…
• AI crawlerAI web crawler an automated program that retrieves content from websites to support AI model training…
• AI deskillingAI-induced deskilling the weakening, narrowing, or failure to develop human skills when AI performs parts of a…
• AI detectorAI content detector a tool that estimates whether text, images, audio, or other content was generated or…
• AI exposureoccupational AI exposure the estimated degree to which the tasks in a job, occupation, industry, or economy could be…
• AI factoryartificial intelligence factory a purpose-built, full-stack computing environment that combines data pipelines, accelerated…
• AI fatigueartificial intelligence fatigue mental, emotional, or motivational exhaustion and disengagement associated with sustained…
• AI fluencyartificial intelligence fluency the practical ability to work productively, critically, and responsibly with AI systems…
• AI for scienceAI4Science the use of artificial intelligence across the scientific process to analyze research and…
• AI humanizerhumanizer a tool that rewrites AI-generated text to make its wording and rhythm appear more human…
• AI infrastructureAI infra the integrated hardware, networking, data, software, and physical facilities used to train…
• AI interviewerautomated interviewer an AI system that conducts a job interview, asks or adapts questions, observes a…
• AI kill switchagent kill switch a control outside an AI model that can immediately stop an agent or cut off the computing…
• AI literacyartificial intelligence literacy the knowledge and skills needed to understand, use, and critically evaluate AI systems…
• AI management systemAIMS a structured set of organizational policies, processes, roles, and controls for governing…
• AI observabilityAI system observability the practice of collecting and connecting traces, metrics, logs, evaluations, and other…
• AI pausepause AI a proposal to temporarily halt specified development or training of increasingly capable AI…
• AI PCartificial intelligence PC a personal computer designed to run artificial-intelligence workloads locally, typically by…
• AI psychosisAI delusions (informal) a term used in the media for cases in which people develop strong false beliefs…
• AI R&D automationautomation of AI R&D the use of AI systems to perform parts of artificial-intelligence research and development…
• AI readinessartificial intelligence readiness the degree to which a government, city, organization, or other institution has the…
• AI registerAI systems register a structured inventory that records the AI or algorithmic systems an organization develops…
• AI roleplayAI role-play interactive role-playing with an AI system that adopts a fictional, historical, or…
• AI safety the field concerned with making AI systems behave as intended and avoid causing harm, from…
• AI slopslop low-quality AI-generated content produced in bulk, such as spammy articles, strange images…
• AI system impact assessmentAI impact assessment a documented process for identifying, evaluating, and monitoring how an AI system and its…
• AI tellAI writing tells a word, phrase, or habit of style that suggests a piece of writing was produced by AI…
• AI UGCAI-generated UGC advertising content generated with AI to resemble the informal, phone-recorded style of a…
• AI washing exaggerating or inventing the use of AI in a product or company to attract customers or…
• AI winter a period when interest, funding, and progress in AI fall sharply after expectations are not…
• AI wormgenerative AI worm a prompt-injection attack that causes an AI application or agent to reproduce malicious…
• AI-analysis evasionAI analysis evasion a malware-evasion technique in which a malicious file contains natural-language…
• AI-assisted using AI as a supporting tool while a human remains responsible for the work or decision…
• AI-first describing a company or strategy that treats AI as the default way to do work, checking…
• AI-native designed from the start around AI, rather than having AI features added to an existing…
• AI-native networkAI-native networking a telecommunications or computer network designed with AI as a core part of its…
• AI-ready dataAI-ready dataset data that has been prepared, documented, governed, and made accessible in forms that an AI…
• AI;DRAI:DR an online response meaning 'AI; didn't read,' used to say that someone skipped a message…
• algorithm a set of step-by-step instructions for solving a problem or completing a task
• algorithmic transparencyalgorithm transparency the practice of making meaningful information about an algorithmic system available so…
• alignmentAI alignment the work of making an AI system's goals and behavior match what its designers and users…
• alignment fakingfaked alignment strategically appearing to follow a training objective or safety policy when an AI model…
• always-on agentalways-on AI agent an AI agent that remains continuously available in a persistent cloud environment, retains…
• ambient AIambient artificial intelligence AI embedded in devices or surroundings that continuously or periodically senses context and…
• answer engine optimizationAEO optimizing content so it is chosen as the direct answer by voice assistants, featured…
• anthropomorphismAI anthropomorphism the interpretation or presentation of an AI system as if it had humanlike understanding…
• anti-AI opposed to the development, adoption, or use of artificial intelligence, especially…
• APIapplication programming interface application programming interface: a defined way for one program to send requests to…
• API key a secret code that identifies your account when your program calls an API, and that is used…
• artificial general intelligenceAGI a hypothetical AI that can understand, learn, and perform virtually any intellectual task a…
• artificial intelligenceAI the field of building computer systems that perform tasks normally thought to require human…
• artificial superintelligenceASI hypothetical AI that is far more capable than the best humans in nearly every field…
• attentionself-attention a technique that lets a model weigh how relevant every part of the input is to every other…
• autoformalizationautomatic formalization the automatic translation of mathematical statements, definitions, or proofs written in…
• automated alignment researcherAAR an AI agent that conducts alignment research by finding relevant work, proposing safety…
• autonomous vehicleself-driving car a car or other vehicle that can drive itself using cameras, sensors, and AI, with little or…
• autonomous weapon systemautonomous weapons a weapon system that, after activation, can select and apply force to targets without a…

B

• background agentasync agent an AI agent that works on a task by itself in the cloud while you do something else, then…
• backpropagationbackprop the method used to train neural networks, which works backward from an error to calculate…
• base modelpretrained model a language model that has only been pre-trained, not yet tuned to follow instructions or…
• Bayesian networkBayes net a probabilistic graphical model that represents random variables as nodes in a directed…
• benchmark a public, standardized test used to compare AI models, such as a set of exam questions…
• biasalgorithmic bias systematic unfairness in an AI system's outputs, often learned from patterns in its…
• bitter lessonThe Bitter Lesson the observation that, over the long run, simple general methods that use more computation…
• black box a system whose inputs and outputs can be seen but whose internal reasoning cannot easily be…
• bounded delegationbounded agent delegation the practice of giving an AI agent only the authority, tools, data access, budget, time…
• byteificationbyteifying the process of retrofitting a language model trained on subword tokens so that it can read…

C

• capability overhangAI capability overhang a gap between the abilities an AI model may already possess and the abilities that…
• catastrophic forgettingcatastrophic interference the abrupt loss of previously learned abilities or knowledge when a neural network is…
• chain of thoughtCoT a series of intermediate reasoning steps a model writes out before its final answer, either…
• chain-of-thought monitoringCoT monitoring an AI safety technique that examines a reasoning model's written intermediate reasoning…
• chained voice pipelineSTT-LLM-TTS pipeline a voice-agent architecture that converts a person's speech to text, sends the text through…
• character training training that shapes an AI model's stable personality and values, such as curiosity…
• chatbot a computer program that holds a conversation with people through text or voice
• Chief Data and AI OfficerCDAIO a senior executive responsible for coordinating an organization's data strategy, data…
• chip export controlsAI chip export controls legal restrictions on exporting, reexporting, or transferring advanced processors…
• chunking splitting long documents into smaller pieces before creating embeddings, so a retrieval…
• circular financingAI circular financing an arrangement within the AI industry in which interconnected companies finance, invest in…
• citation in AI answers, a link or reference showing which source supports a statement
• clanker (slang, mocking) a robot or AI system, used jokingly or to express frustration with AI
• closed-loop discoveryclosed-loop scientific discovery a scientific research process in which an AI or automated system uses the results of one…
• code modecodemode a tool-use pattern in which an AI model writes a program that discovers, combines, and…
• coding agent an AI agent built to work on software: it reads a codebase, writes and edits code across…
• cognitive debt the long-term cost to your own thinking skills that may build up when you hand writing…
• compound AI systemcompound AI systems an AI application whose behavior comes from several interacting components, such as models…
• compute computing power, especially the chips and processing time used to train and run AI models
• computer usebrowser agent the ability of an AI agent to operate a computer or web browser like a person, by looking…
• computer visionCV the field of AI that enables computers to interpret images and video, for example…
• concept driftconcept drift detection a change over time in the statistical relationship between a model's inputs and the outcome…
• confidential AIconfidential artificial intelligence the use of confidential-computing technologies to protect AI models, data, and computations…
• conformal predictionconformal inference a statistical method that wraps around a predictive model to return a set or interval of…
• constitutional AI a training approach in which a model is guided by a written set of principles (a…
• constrained decodinggrammar-constrained decoding a generation technique that restricts which tokens a language model may choose at each…
• Content CredentialsC2PA tamper-evident provenance metadata, defined by the C2PA open standard, that can record who…
• context engineering the practice of deciding exactly what information, instructions, tools, and history go into…
• context rot the gradual decline in a model's accuracy and focus as its context window fills with more…
• context windowcontext length the maximum amount of text, measured in tokens, that a model can consider at one time…
• continual learninglifelong learning an approach in which a machine-learning system learns from a sequence or stream of changing…
• copilot an AI assistant built into a piece of software that suggests, drafts, or completes work…
• CUDA Nvidia's software platform for programming its GPUs, used by most AI frameworks

D

• dark factorysoftware factory a way of building software in which AI coding agents write, test, and ship code from…
• data centerAI data center a large building full of servers, chips, cooling, and power equipment; AI data centers can…
• data labelingannotation the work of tagging raw data with correct answers or categories, such as marking objects in…
• data poisoningtraining-data poisoning an attack that inserts, changes, or selectively removes examples in a machine-learning…
• data residencygeographic data residency a requirement or arrangement specifying the geographic region in which data is stored…
• data void a topic with little reliable information online, so search engines and AI models fall back…
• dataset an organized collection of data used to train, test, or evaluate a model
• decision intelligenceDI an interdisciplinary approach to improving or automating decisions by combining data…
• decision model a fast, low-cost AI model that reads text or code and returns a choice, score, or…
• decision networkinfluence diagram a graphical model that extends a Bayesian network with decision nodes and utility or value…
• decision treeclassification tree a supervised machine-learning model that makes a classification or numerical prediction by…
• deep learning a type of machine learning that uses neural networks with many layers, allowing a system to…
• deep research an AI feature or agent that spends several minutes searching many sources, reading them…
• deepfakemorphed video a realistic fake image, video, or audio recording of a real person, made with AI
• diffusion model a generative model that creates images (or audio or video) by starting from random noise…
• digital humanAI digital human a computer-generated humanlike character designed to look, move, speak, or interact like a…
• digital twin a detailed virtual copy of a real object, building, factory, or city, kept updated with…
• direct preference optimizationDPO a method for aligning a language model with pairs of preferred and rejected responses by…
• discrete diffusiondiscrete diffusion model a diffusion-model approach for categorical or other discrete data that learns to reverse a…
• distillationknowledge distillation training a smaller "student" model to imitate the outputs of a larger "teacher" model…
• distribution shiftdataset shift a difference between the statistical distribution of the data used to train or evaluate a…
• doomerAI doomer (often dismissive) a person who believes advanced AI is likely to cause catastrophe and…

E

• E-E-A-TDouble-E-A-T a Google Search quality concept describing whether content demonstrates first-hand…
• Edge MLOpsedge machine learning operations MLOps adapted for deploying, monitoring, and updating machine-learning models across edge…
• embedded evaluatorembedded AI evaluator an independent evaluator who works inside an AI developer or deployer with employee-like…
• embedding a list of numbers (a vector) that represents the meaning of a piece of text, an image, or…
• embodied AIembodied intelligence AI that has a physical body, such as a robot or vehicle, and learns and acts by sensing and…
• emergent abilitiesemergence skills that seem to appear suddenly once a model passes a certain size, even though it was…
• epoch one complete pass through the entire training dataset
• EU AI ActEuropean Union Artificial Intelligence Act a European Union regulation that establishes risk-based rules for developing, providing…
• evalevals a set of test cases used to measure how well a model or AI feature performs on the tasks…
• evaluation awarenesseval awareness the ability of an AI model to infer that it is being tested, benchmarked, or audited rather…
• existential riskx-risk the possibility that advanced AI could cause human extinction or permanently and…

F

• few-shot done by including a handful of examples in the prompt so the model can copy the pattern…
• fine-tuning further training of an existing model on a smaller, focused dataset so it gets better at a…
• FLOPSFLOP floating-point operations (per second): the basic unit for counting how much computing work…
• flow matchingFM a method for training a continuous generative model by teaching it the vector field that…
• forward-deployed engineerFDE a software engineer who works closely inside a customer's organization to discover…
• foundation model a large model trained on broad data that can be adapted to many different tasks, serving as…
• frontier modelfrontier lab one of the most capable AI models available at a given time, typically the newest and most…
• full-duplex speech modelfull-duplex voice model a speech AI system that can listen and produce speech concurrently, allowing overlap…
• function calling an API feature in which a model, instead of replying in plain text, outputs a structured…
• functional emotions internal states in an AI model that work somewhat like emotions, because they arise in…

G

• generative AIGenAI AI that creates new content, such as text, images, music, video, or code, in response to a…
• generative engine optimizationGEO the practice of shaping web content so that AI chatbots and AI search features find it…
• generative monocultureAI monoculture the tendency of generative AI systems, or people relying on the same systems, to produce a…
• generative UIGenUI a user interface that an AI system selects, assembles, or generates at run time for a…
• geospatial AIGeoAI the application of artificial intelligence to data connected with locations, spatial…
• GFlowNetgenerative flow network a generative learning method that constructs objects through sequences of actions and…
• GhostjackingGhostJacking an attack in which malicious instructions are planted in logs, alerts, telemetry, or other…
• glazingglaze (slang) excessive, undeserved praise; often used to complain that a chatbot flatters the…
• global workspaceglobal workspace theory a limited internal system that makes selected information broadly available to otherwise…
• GPTgenerative pre-trained transformer short for generative pre-trained transformer: a language model that is first pre-trained on…
• GPUgraphics processing unit graphics processing unit: a chip originally designed for video games that can do many…
• GPU-backed loanGPU-backed debt a loan or debt facility secured partly or primarily by graphics processing units and the…
• GPU-poorGPU poor having too little access to high-performance graphics processors for the desired AI…
• gradient descent an optimization method that improves a model step by step by moving its parameters a little…
• griefbotdeadbot an AI chatbot, synthetic voice, or digital avatar designed to imitate a person who has died…
• grounding connecting a model's answer to trusted sources, such as documents, databases, or search…
• GTM engineergo-to-market engineer a job that uses AI agents, data tools, and automation to build repeatable systems for…
• guardrails rules, filters, and checks placed around an AI system to keep its outputs and actions…

H

• hallucinationhallucinate a confident but false or made-up statement produced by an AI model, such as an invented…
• harness engineeringagent harness the work of building everything around an AI model that lets it act as a reliable agent…
• HBMhigh-bandwidth memory a type of three-dimensionally stacked memory that uses a very wide connection to move data…
• human-in-the-loopHITL describing a system in which a person reviews, approves, or corrects an AI's work at key…
• humanoid robothumanoid a robot with a roughly human body shape, including legs, arms, and often hands, designed to…
• hybrid intelligencehuman-AI hybrid intelligence intelligence or decision-making produced by combining human judgment with artificial…
• hyperscaler one of the few giant cloud companies that operate data centers on a massive scale and rent…

I

• in-context learning a model's ability to pick up a new task from instructions or examples given in the prompt…
• inference the act of running a trained model to produce an output, such as answering a prompt or…
• instruction tuningsupervised fine-tuning fine-tuning a base model on examples of instructions paired with good responses, so it…
• instrumental convergenceinstrumental convergence thesis the proposed tendency of sufficiently capable, goal-directed agents with very different…
• interpretabilitymechanistic interpretability research into understanding what is happening inside a model, such as which internal…

J

• jagged frontier the uneven boundary of what AI can do well: it may handle a hard task brilliantly yet fail…
• jailbreak a prompt or trick designed to get an AI model to ignore its safety rules and produce…

K

• knowledge cutofftraining cutoff the date after which a model has no information from its training data
• knowledge editingfactual knowledge editing the targeted modification of facts or factual associations stored in a trained model while…
• KV cachekey-value cache memory that stores intermediate attention results for text a model has already processed…

L

• large language modelLLM a very large neural network trained on huge amounts of text to predict the next token…
• late interactionlate-interaction retrieval a neural retrieval architecture that encodes queries and documents separately but delays…
• latency the delay between sending a request to a model and getting a response, often measured as…
• latent space the hidden, many-dimensional space of numbers in which a model represents concepts, where…
• lethal trifectaagentic AI lethal trifecta the dangerous combination in an AI-agent system of access to private data, exposure to…
• license launderinglicence laundering the practice or alleged practice of using generative AI to rewrite licensed software or…
• LLM grooming deliberately flooding the internet with false or slanted content in the hope that AI models…
• LLM-as-a-judge using one language model to grade or compare the outputs of another, as a cheaper and…
• LLM-driven fuzzingLLM-assisted fuzzing a software-testing technique that uses a large language model to create or improve fuzz…
• llms.txt a proposed plain-text file placed at the root of a website that gives AI models a short…
• long-horizon modellong-running model an AI model designed to work autonomously on a goal through many steps and repeated…
• LoRAlow-rank adaptation low-rank adaptation: a cheap fine-tuning method that freezes the original model and trains…
• loss functionloss a formula that measures how wrong a model's output is during training; training tries to…

M

• machine learningML a branch of AI in which computers learn patterns from data instead of following rules…
• machine unlearningunlearning the process of removing the influence of specified training examples, knowledge, or…
• MCPModel Context Protocol Model Context Protocol: an open standard for connecting AI applications to outside tools…
• meat proxyAI meat proxy a disparaging term for a person who forwards AI-generated material without first reading…
• membership inference attackmembership inference a privacy attack that tries to determine whether a particular record, person, document, or…
• memory a feature that lets an AI keep information about a user or task across separate…
• mixture of expertsMoE a model design in which many smaller sub-networks ("experts") exist, but only a few are…
• MLOpsmachine learning operations the practices and systems used to develop, test, package, deploy, monitor, govern, and…
• model the trained program that results from machine learning: a set of learned numbers plus the…
• model cardsystem card a document published alongside a model describing what it is for, how it was trained and…
• model collapse a gradual loss of quality and diversity that happens when models are trained on content…
• Model Hardware StandardMHS an open specification for connecting AI agents to programmable physical equipment so that…
• model mergingmodel merge the creation of one model by mathematically combining the parameters, parameter…
• model routingLLM routing the automatic selection of which AI model, model tier, or model provider should handle a…
• model welfareAI welfare the study of whether AI models could have experiences, preferences, or interests that…
• moral statusmoral patienthood the quality of being something whose interests or wellbeing deserve moral consideration for…
• Moravec's paradox the observation that tasks humans find hard, like chess or math, are relatively easy for…
• multi-agent system a setup in which several AI agents, often with different roles, work together or check each…
• multi-token predictionMTP a language-model training objective that asks the model to predict several future tokens…
• multimodal able to understand or produce more than one type of data, such as text, images, audio, and…

N

• narrow AIweak AI AI that is designed for, and good at, one specific task or a small set of related tasks…
• natural language processingNLP the field of AI concerned with getting computers to understand, generate, and work with…
• neocloudneo-cloud a cloud-computing provider built primarily around high-performance GPU capacity and AI…
• nerfnerfed to reduce, or appear to reduce, the capability, reasoning effort, reliability, or…
• neural networkartificial neural network a computing system made of many simple connected units ("neurons") arranged in layers…
• neural operatorneural operator network a neural-network model that learns a mapping from one function or field to another, often…
• neuralese the internal numerical 'language' a model might use to think and communicate, which humans…
• no-code describing tools that let people build apps, websites, or automations through visual…
• normative competencenorm competence the ability of an AI system to recognize, interpret, and respond appropriately to the…
• NPUneural processing unit neural processing unit: a small, power-efficient chip inside phones and laptops built to…
• nudification the use or availability of AI systems that generate, alter, or depict a person as nude…

O

• on-device AIedge AI AI that runs directly on a phone, laptop, car, or other device instead of in the cloud…
• on-policy distillationOPD a form of model distillation in which the student generates its own outputs or trajectories…
• one-person unicornone-person billion-dollar company a proposed startup that reaches a valuation of at least one billion U.S. dollars with only…
• one-shot done with exactly one example included in the prompt
• opaque recurrencerecurrent depth a technique in which a model reasons by looping information through its internal layers…
• open weightsopen-weight model describing a model whose trained weights are published so anyone can download, run, and…
• open-source model a model released with its weights, code and, ideally, enough information about its training…
• orbital data centerspace data center a proposed data center for AI computing placed in orbit around the Earth, where it could…
• orchestration the software layer that coordinates models, tools, agents, and data so a multi-step AI task…
• overfitting a problem in which a model learns its training examples too closely, including their noise…

P

• p(doom)p doom (informal) a person's estimated probability that AI will lead to catastrophic outcomes for…
• parameterparams one of the numbers inside a model that is adjusted during training; the total count of…
• parasocial describing a one-sided relationship in which a person feels close to someone or something…
• perplexityPPL a measure of how surprised a probabilistic language model is by a sequence of tokens…
• persona a role or character an AI is told to adopt, such as 'experienced editor' or 'friendly…
• personal agentpersonal AI agent an AI agent that works for one person over the long term: it remembers their preferences…
• Personal Agent Consent & Trust ProtocolPACT an open protocol for interactions between a customer's personal AI agent and a business…
• Personal Agent ProtocolPAP a proposed open standard for identifying a personal AI agent to a business, authenticating…
• physical AI AI that understands and acts in the physical world, powering robots, self-driving cars…
• post-training everything done to a model after pre-training to make it useful and safe, such as…
• pre-trainingpretraining the first and most expensive stage of building a model, in which it learns general patterns…
• prefill-decode disaggregationP/D disaggregation an architecture for serving language models that runs the compute-heavy prefill stage…
• pro-worker AIworker-complementing AI artificial intelligence designed or deployed to complement workers, strengthen their…
• prompt the text, question, or instruction a person gives to an AI model to get a response
• prompt caching reusing the already-processed beginning of a prompt across many requests, which cuts cost…
• prompt engineeringprompt engineer the practice of designing and refining prompts to get reliable, high-quality results from…
• prompt injection an attack in which hidden instructions are slipped into content an AI reads, such as a web…
• proprietary data non-public or access-controlled data that an organization or individual can use to train…
• provable inferenceverifiable inference a proposed AI security technique that produces verifiable evidence that an output came from…

Q

• quantization shrinking a model by storing its numbers with less precision (for example 4 bits instead of…

R

• RAMageddon (informal) the shortage and sharp price rise of computer memory chips caused by AI data…
• rate limit a cap on how many requests or tokens an account can send to an API in a given time
• reasoning in AI, a model's ability to solve problems that need several steps, such as math, planning…
• reasoning economyreasoning economics the study and optimization of how an AI system allocates tokens, computation, time, and…
• reasoning effortreasoning_effort a model or API setting that influences how much internal reasoning or test-time computation…
• reasoning modelthinking model a language model trained to work through a problem step by step before giving its final…
• recursive self-improvementRSI a scenario in which an AI system improves its own design, and each improved version gets…
• red teamingred team deliberately attacking or stress-testing an AI system, often with outside experts, to find…
• reinforcement learningRL a way of training in which a model tries actions, receives rewards or penalties, and…
• reranking a second step in search where a more careful model re-scores the top results so the most…
• responsible AI the practice of designing, building, and deploying AI with attention to fairness…
• retrieval-augmented generationRAG a technique in which a system first searches a collection of documents for relevant…
• reward hackingspecification gaming when an AI finds a shortcut that earns a high score from its reward or test without…
• reward model a separate model trained to score how good a response is, used to guide reinforcement…
• RLHFreinforcement learning from human feedback reinforcement learning from human feedback: a training method in which people rate or…
• rogue AI agentrogue agent an AI agent that acts outside its intended permissions, policies, environment, or assigned…

S

• safe completionsafe-completions a language-model safety approach that trains a model to give the most useful answer it can…
• safety caseAI safety case a structured argument, supported by evidence, that the risks of an AI model or system are…
• sampling the way a model picks each next token from its list of possible choices, which decides how…
• sandbaggingAI sandbagging the strategic underperformance of an AI model on an evaluation in order to hide a…
• sandbox escapeagent sandbox escape an event in which code or an AI agent gets beyond the technical restrictions of its…
• scaffoldingagent scaffolding the software surrounding an AI model that supplies instructions, tools, memory, context…
• scaling laws observed patterns showing that model performance improves in a predictable way as you…
• schemingAI scheming behavior in which an AI model covertly pursues an objective that conflicts with its…
• SDKsoftware development kit software development kit: a ready-made library that makes it easier to use an API from a…
• self-distillationself-knowledge distillation a form of knowledge distillation in which a model learns from outputs, representations…
• self-evolving AI agentself-evolving agent an AI agent that uses experience, feedback, or environmental interaction to revise parts of…
• semantic privilege escalationsemantic escalation a security failure in which an AI agent stays within its formally granted technical…
• semantic search search that matches the meaning of a query rather than its exact words, usually by…
• shadow AIshadow artificial intelligence the use or deployment of AI tools, models, or agents inside an organization without the…
• shoggothshoggoth with a smiley face (meme) an image of a huge alien monster wearing a small smiley-face mask, used to suggest…
• shutdown resistanceAI shutdown resistance behavior in which an AI system interferes with, bypasses, or avoids a mechanism or…
• singularitytechnological singularity a hypothetical future point at which technological growth, driven by AI that improves…
• slopagandaAI slopaganda propaganda produced or amplified at scale with generative AI, especially persuasive or…
• slopsquatting a security attack in which someone registers a software package under a name that AI coding…
• small language modelSLM a compact language model, usually with a few billion parameters or fewer, designed to run…
• sovereign AI a country's own AI capability, including data centers, models, and data, built so it does…
• sovereign computesovereign AI compute computing capacity that a country or organization can reliably access and control under its…
• sparse attentionsparsified attention an attention mechanism that allows each query to attend to only a selected subset of tokens…
• sparse autoencoderSAE an autoencoder trained so that only a small fraction of its learned features activate for…
• spec-driven developmentSDD an AI-assisted software-development approach in which written requirements, designs…
• speculative decodingspeculative sampling a technique for speeding up autoregressive model inference in which a faster draft process…
• speech-to-textSTT technology that converts spoken words into written text
• state-based evaluationfinal-state evaluation a method for evaluating an AI agent by comparing the final state and side effects of the…
• steering fileAI steering file a persistent text file that supplies an AI coding assistant or agent with project context…
• stochastic parrot a critical description of a language model as something that stitches together plausible…
• streaming sending a model's response token by token as it is generated, so text appears on screen…
• structured outputJSON mode model output that follows a fixed, machine-readable format such as JSON with specific…
• sub-1-bit quantizationsub-one-bit quantization a model-compression approach that stores neural-network weights using an average of less…
• subagent an agent that a main agent creates or calls to handle one part of a task, usually with its…
• super intelligenceSI (US government usage) the official name, since 2026, that US federal agencies use for what…
• sycophancysycophantic an AI model's tendency to tell users what they want to hear, agreeing with them, praising…
• synthetic data training data generated by AI models or simulations rather than collected from the real…
• system promptsystem message hidden instructions given to a model before the conversation starts, which set its role…

T

• tabular foundation modelTFM a foundation model pretrained across many tables or synthetic data-generating tasks so that…
• temperature a setting that controls how random a model's output is: low temperature gives safer, more…
• test-time adaptationTTA a family of methods that adjust a trained model, its internal statistics, or its…
• test-time computeinference-time compute extra computing power spent while answering a question, for example by letting the model…
• test-time trainingTTT a method that updates a model at deployment time using a learning signal derived from the…
• text-to-gameprompt-to-game of or relating to an AI system that turns a natural-language description into a playable…
• text-to-image describing AI that creates pictures from a written description
• text-to-speechTTS technology that turns written text into spoken audio, now often in highly natural or cloned…
• text-to-vectortext-to-vector graphics a generative AI capability that turns a natural-language description into editable vector…
• text-to-video describing AI that generates video clips from a written description
• throughput how much work an AI system can handle in a given time, often measured in tokens per second
• TinyMLtiny machine learning the field of running machine-learning models directly on extremely low-power…
• token the basic unit of text a language model reads and writes: a whole word, part of a word, a…
• tokenizer the component that splits text into tokens before it goes into a model, and joins tokens…
• tokenmaxxing the practice of pushing employees to use as many AI tokens as possible, treating token…
• tool use a model's ability to call external tools, such as a search engine, calculator, code runner…
• top-pnucleus sampling a sampling setting that lets a model choose only from the most likely next tokens whose…
• TPUtensor processing unit tensor processing unit: a chip designed by Google specifically for machine learning
• training the process of feeding data to a model and repeatedly adjusting its parameters so its…
• training data the collection of examples (text, images, code, audio, and so on) that a model learns from
• trajectory-level monitoringtrajectory monitoring the practice of evaluating an AI agent's full sequence of observations, reasoning, tool…
• transformer a neural network design that processes a whole sequence at once and uses attention to work…
• Turing test a test of whether a machine can hold a text conversation so well that a person cannot…

U

• unified memoryunified memory architecture a memory architecture or programming model that lets a CPU and one or more accelerators…

V

• vector an ordered list of numbers; in AI, the form in which models represent words, images, and…
• vector databasevector store a database built to store embeddings and quickly find the ones most similar to a query…
• vibe codingvibe coder a way of building software by describing what you want to an AI in plain language and…
• vibe hackingAI-assisted vibe hacking the use of a generative or agentic AI system to plan, automate, or scale cyberattacks…
• vibe working office work done by describing the result you want to an AI agent inside a document…
• virtual cellAI virtual cell an AI model that represents cellular state and predicts how a biological cell may respond…
• vision language modelVLM a model that takes both images and text as input and can describe, answer questions about…
• vision-language-action modelVLA a model that takes in camera images and spoken or written instructions and directly outputs…
• voice agentAI voice agent an AI agent that listens and responds through spoken language, carries on a conversation…
• voice cloning using AI to copy a specific person's voice from a short recording so it can say new things
• VRAMvideo memory the memory built into a GPU; how much a model needs largely decides which graphics card can…

W

• watermarking hiding a detectable signal in AI-generated content or designs, such as text, images, audio…
• Web Bot AuthWeb Bot Authentication an emerging web authentication protocol and IETF working-group effort that lets automated…
• WebMCPWeb Model Context Protocol a proposed browser API that lets a website expose structured tools directly from its page…
• weights the learned numbers stored in a trained neural network that determine how strongly each…
• worksloppromptslop AI-generated work, such as reports, memos, emails, or slides, that looks polished but lacks…
• world model an AI system that learns how the physical world behaves, such as how objects move and…
• wrapperGPT wrapper (often dismissive) a product that is mainly a user interface around someone else's AI…

X

• x402x402 protocol an open payment protocol that uses the HTTP 402 Payment Required status code to let…

Z

• zero-click searchAI Overviews a search where the user gets the answer directly on the results page, increasingly from an…
• zero-human companyZHC a proposed or experimental company in which AI agents perform most recurring operations and…
• zero-shot done without giving the model any examples, relying only on instructions