You are here
Current learning moveOrient
Find an entrance using an ordinary question while seeing the role of every result.
inAiLoading page content...
…
inAiFind an entrance using an ordinary question while seeing the role of every result.
inAiLoading page content...
Universal search
Search returns objects, not just pages. Results show format, depth, time, and provenance so you can choose the right kind of answer.
All active learning objects
Guided learning
Paths, units, and lessons designed to be followed in an order.Eight active segments build an ordinary-language map of AI, practical judgment rules, and a learner-chosen next direction.
orient · understand · try · 30 min · inai-originalWhy capability, accessible interfaces, infrastructure, investment, and adoption made an older field feel suddenly new.
orient · 4 min · inai-explanationA practical definition based on capabilities, learned or designed behavior, and the wider product around a model.
orient · understand · 4 min · inai-explanationA field map separating perception, prediction, generation, and action while showing how products combine them.
orient · understand · 4 min · inai-explanationA beginner distinction between training parameters from examples and running a trained model on a new input.
understand · 4 min · inai-explanationA practical separation of plausibility, support, freshness, and stakes for generated output.
understand · try · 5 min · inai-explanationA comparison of model, chatbot, workflow, and agent through goals, tools, state, permissions, approvals, and recovery.
orient · understand · 4 min · inai-explanationA sharing boundary based on authorization, sensitivity, provider behavior, minimization, and consequences.
orient · try · 3 min · inai-explanationA transparent choice among deeper explanation, experimentation, practice, and a substantial external course.
orient · 2 min · inai-originalA compact first route from generated output to practical verification through four connected learning objects.
orient · understand · try · build · 40 min · inai-originalA useful model of generation, context, support, recency, and stakes—followed by active judgment and transfer work.
understand · learn · 12 min · inai-explanationBuild enough connected understanding to follow ordinary AI debate, inspect systems and claims, make practical privacy and verification decisions, and keep learning without depending on headlines, marketing, or fear.
understand · learn · try · examine · 310 min · inai-originalThe field is old; the present combination of capability, scale, interface, infrastructure, investment, and deployment is new. Good judgment begins by keeping both facts visible.
understand · learn · try · examine · 25 min · inai-explanationAI is easier to understand as overlapping capabilities, methods, system layers, and application domains than as one ladder from simple to intelligent.
understand · learn · try · examine · 30 min · inai-explanationLearned behavior reflects examples, representation, objectives, feedback, evaluation, and deployment—not raw data acting by itself.
understand · learn · try · examine · 30 min · inai-explanationA trained model is one component. Development, evaluation, inference, integration, operation, monitoring, and human use create the behavior and consequences people encounter.
understand · learn · try · examine · 25 min · inai-explanationGenerative models produce new sequences or media from learned patterns and current context. This explains fluency and variation without turning output into memory, intention, or verified knowledge.
understand · learn · try · examine · 35 min · inai-explanationAn output can change because the request, available context, retrieved evidence, tool access, system instructions, or workflow changed. These layers improve capability only within their own quality and permission limits.
understand · learn · try · examine · 30 min · inai-explanationDo not ask whether AI is trustworthy in the abstract. Identify the claim, support, recency, uncertainty, decision, affected people, and cost of error, then choose a proportionate verification route.
understand · learn · try · examine · 35 min · inai-explanationA useful system does not cancel confidentiality, privacy, security, professional duty, or responsibility. Control begins before data leaves your hands and continues through every action and consequence.
understand · learn · try · examine · 30 min · inai-explanationAgent-like systems continue across steps toward a goal and may select tools or actions. The important questions are authority, state, observation, approval, failure, and recovery—not a human-sounding identity.
understand · learn · try · examine · 30 min · inai-explanationAI changes through institutions, work, education, media, infrastructure, law, markets, and public choices. Futures deserve evidence and plural viewpoints—not inevitability, panic, or a single intelligence score.
understand · learn · try · examine · 40 min · inai-explanationRecognize the AI already around you, choose tools for a real purpose, ask for useful help, verify important output, protect ordinary privacy, and keep responsibility for the final decision.
understand · try · use · 165 min · inai-originalAI may rank a feed, detect fraud, predict a route, transcribe speech, recommend a product, or generate an image. Learning to identify the capability and the surrounding system is more useful than guessing from a shiny interface.
understand · try · use · 25 min · inai-explanationThe best-known or most capable tool in a benchmark is not automatically the right tool for your task. A useful choice connects purpose, evidence, data conditions, total cost, accessibility, and failure consequences.
understand · try · use · 30 min · inai-explanationA useful request makes the task legible: purpose, relevant context, constraints, evidence needs, and output form. It improves the conditions for an answer but cannot guarantee truth, judgment, or safe use.
understand · try · use · 30 min · inai-explanationNot every output needs an investigation. Verification should rise with consequence, uncertainty, novelty, and irreversibility—and should examine the underlying claim rather than ask the same generator to reassure itself.
understand · try · use · 30 min · inai-explanationPrivacy decisions happen before you paste, upload, connect, record, or authorize. The practical question is not only whether information feels secret, but whether it is necessary, permitted, identifiable, retained, and exposed to consequence.
understand · try · use · 25 min · inai-explanationAI can expand options, reduce friction, and help inspect a problem. It cannot inherit your relationships, duties, values, or accountability merely because its response is fluent or convenient.
understand · try · use · 25 min · inai-explanationAnalyze work at the level of tasks, responsibilities, relationships, and workflows; distinguish automation from augmentation; place review according to stakes; and prepare concrete worker–manager conversations.
understand · try · use · examine · 180 min · inai-originalA job title contains many tasks, responsibilities, relationships, constraints, and forms of judgment. An AI capability may change some of them without determining the future of the whole job.
understand · try · use · examine · 30 min · inai-explanationAI may replace a step, accelerate it, change how a person performs it, create new work, or do several at once. The useful question is how the work system changes and for whom.
understand · try · use · examine · 30 min · inai-explanationAdding AI to an unmapped process can accelerate confusion and move failures downstream. A redesign starts with the purpose, people, handoffs, evidence, exceptions, and controls already present.
understand · try · use · examine · 30 min · inai-explanationHuman oversight is not a person placed somewhere after an AI output. It is a designed ability to understand evidence, intervene before consequence, identify ownership, and recover when the system fails.
understand · try · use · examine · 30 min · inai-explanationAI does not produce one universal future-skills list. It often increases the value of domain judgment, clear communication, verification, workflow understanding, and the ability to use and challenge systems responsibly.
understand · try · use · examine · 30 min · inai-explanationUseful workplace dialogue moves beyond ‘Will AI replace us?’ to specific questions about purpose, tasks, data, monitoring, evidence, workload, expectations, support, responsibility, and recourse.
understand · try · use · examine · 30 min · inai-explanationLearn to inspect what an agentic system actually does: how it continues work, uses software, carries state, receives authority, and recovers when action goes wrong.
understand · try · use · examine · 165 min · inai-originalClassify systems without relying on marketing labels
understand · try · use · examine · 25 min · inai-explanationUnderstand continuation across work
understand · try · use · examine · 25 min · inai-explanationExplain how agents act through software
understand · try · use · examine · 25 min · inai-explanationDistinguish temporary context from durable state
understand · try · use · examine · 30 min · inai-explanationEvaluate what an agent may access and change
understand · try · use · examine · 30 min · inai-explanationDesign approval and recovery for consequential action
understand · try · use · examine · 30 min · inai-explanationBuild an evidence practice for generated claims and synthetic media, then connect individual verification to bias, privacy, institutions, rights, and public oversight.
understand · try · use · examine · 170 min · inai-originalRecognize fluent unsupported output
understand · try · use · examine · 25 min · inai-explanationTrace a claim beyond generated text
understand · try · use · examine · 30 min · inai-explanationEvaluate evidence in a synthetic-media environment
understand · try · use · examine · 30 min · inai-explanationLocate bias across data, objective, deployment, and institution
understand · try · use · examine · 30 min · inai-explanationDistinguish personal convenience from systemic data risk
understand · try · use · examine · 25 min · inai-explanationUnderstand why context, rights, risk, and oversight matter
understand · try · use · examine · 30 min · inai-explanationLearn to use AI for explanation, feedback, exploration, and creation while preserving the thinking, practice, authorship, and shared judgment that make learning and creative work meaningful.
understand · try · use · examine · 165 min · inai-originalThe same tool can ask a useful question, explain a difficult idea, suggest a direction, or silently complete the work. The learning effect depends on the purpose, the learner’s action, and what happens after the output appears.
understand · try · use · examine · 25 min · inai-explanationAccess to generated explanations changes which supports are available, not the need to build knowledge, methods, values, and judgment. The useful question is which capabilities a person must possess to notice, question, connect, and act without blind dependence.
understand · try · use · examine · 25 min · inai-explanationA polished product is weaker evidence when assistance is easy to obtain and hard to observe. Assessment can respond by collecting several kinds of evidence across process, explanation, application, sources, decisions, and revision.
understand · try · use · examine · 30 min · inai-explanationAI can expand variation and lower the cost of drafts, but creative judgment still concerns purpose, selection, transformation, context, attribution, rights, and responsibility. A useful account names what each participant and system contributed.
understand · try · use · examine · 30 min · inai-explanationDependence is not simply frequent tool use. The practical question is whether a person’s ability to begin, judge, recover, explain, and choose grows or shrinks when the support is removed, wrong, or unavailable.
understand · try · use · examine · 30 min · inai-explanationA useful agreement turns abstract concern into shared decisions about purposes, privacy, learning evidence, disclosure, human support, and revision. It is written with affected people and tested against realistic situations.
understand · try · use · examine · 25 min · inai-explanationExplanations
Topics and quick answers for understanding one question.A connected topic environment for understanding useful output, unsupported claims, evidence, recency, and stakes.
orient · understand · try · examine · build · 35 min · inai-explanationA two-minute distinction between fluent usefulness, factual support, current information, and consequences.
orient · 2 min · inai-explanationArtificial intelligence refers to machine-based systems that infer from inputs how to produce outputs such as predictions, content, recommendations, or decisions. It is a broad field, not one model or product.
orient · 4 min · inai-explanationAI is decades old. Recent attention follows advances in learning methods, computing, data, investment, and systems that can generate useful text, images, audio, and code.
orient · 4 min · inai-explanationMachine learning is a branch of AI in which a model’s behavior is fitted from data, examples, or feedback instead of specifying every rule by hand.
orient · 4 min · inai-explanationGenerative AI creates text, images, audio, video, code, or other content by producing likely structures learned during training and conditioned by current input.
orient · 4 min · inai-explanationAn AI model is a computational structure whose parameters encode patterns fitted during training. Given an input, it produces an estimate, classification, ranking, or generated continuation.
orient · 3 min · inai-explanationDuring training, a system adjusts model parameters using data, objectives, and feedback. During inference, the trained model receives a new input and produces an output.
orient · 4 min · inai-explanationA prompt is text, an image, audio, code, or other input supplied to a generative system. It can state a task, context, constraints, examples, and desired format.
orient · 3 min · inai-explanationContext includes the current prompt and may include system instructions, earlier messages, uploaded material, retrieved passages, tool results, metadata, or state supplied to the model.
orient · 4 min · inai-explanationRetrieval finds relevant material from a database, document set, search system, or tool and places it in the task context. Grounding means constraining or supporting an output with specified evidence or environment.
orient · 5 min · inai-explanationAn AI agent is commonly understood as a system that pursues a goal across multiple steps, maintains state, and may choose or use tools to observe or change an environment.
orient · 5 min · inai-explanationA chatbot may answer one turn without pursuing a goal or acting. An agent-like system continues across steps, maintains state, and may choose tools or actions.
orient · 3 min · inai-explanationLanguage models represent and manipulate relationships well enough to answer, translate, and reason in many tasks. That functional success is real.
orient · 5 min · inai-explanationDo not paste passwords, secret keys, private records, confidential work, protected health or financial data, child data, or another person’s information unless the exact use is authorized and protected.
orient · 5 min · inai-explanationAI can automate or accelerate parts of work, and some roles may shrink, grow, or change. A job also contains relationships, exceptions, responsibility, physical work, and organizational context.
orient · 5 min · inai-explanationDeepfakes use AI to create or modify image, audio, or video so someone appears to say or do something. Synthetic media can be deceptive, disclosed, artistic, or assistive.
orient · 4 min · inai-explanationAI behavior reflects data selection, labels, objectives, model design, thresholds, deployment, institutions, and feedback. Some measurable disparities can be reduced.
orient · 5 min · inai-explanationAn open model may release weights, code, architecture, data information, or documentation so others can inspect, run, modify, or redistribute some parts.
orient · 5 min · inai-explanationArtificial general intelligence usually means AI with broad, adaptable competence across many domains rather than one narrow task. There is no agreed test, threshold, architecture, or timeline.
orient · 6 min · inai-explanationWrite the claim precisely: capability, product version, population, comparison, date, and condition. Then find the original source and inspect method, scope, uncertainty, incentives, and contrary evidence.
orient · 7 min · inai-explanationShare only information that is necessary, authorized, minimized, and suitable for the exact service and consequence.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAn agent-like system continues toward a goal through state, steps, tools, and bounded action.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAI changes tasks before it changes whole job titles, and organizations determine how those changes are distributed.
orient · understand · learn · try · examine · build · 35 min · inai-explanationTrust media through source, provenance, context, and corroboration—not realism or one detector.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAI changes how students can write, solve, explain, practise, and receive feedback, but it does not remove the need to decide what school should help them learn.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAI can remove routine effort, extend access, and support creation, or it can replace the attempts, memory, judgment, and practice through which people remain capable.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAI depends on chips, data centres, networks, electricity, cooling, water, materials, construction, and labour, while some AI applications may also support parts of the energy system.
orient · understand · learn · try · examine · build · 35 min · inai-explanationFairness is a system and institutional question, not a property proved by removing one field or reporting one metric.
orient · understand · learn · try · examine · build · 35 min · inai-explanationAGI is a disputed label for broad, adaptable AI capability; definition, system boundary, test, and timeline remain unsettled.
orient · understand · learn · try · examine · build · 35 min · inai-explanationApply
Labs and Practice tools for testing an idea or producing a decision.A deterministic prediction-before-reveal Lab showing how context and reliable source support change an answer.
try · 15 min · inai-originalAn eight-question practice tool that produces a local, printable verification checklist without a score.
build · 8 min · inai-originalCompare convincing claim formats without mistaking confidence, prestige, or repetition for support.
try · 18 min · inai-originalClassify realistic information by sensitivity, provider context, purpose, and consequence instead of applying one universal rule.
try · 15 min · inai-originalAssemble capability, context, tools, state, permissions, and review to see why a functioning system is more than its model.
try · 20 min · inai-originalSet access, limits, approvals, logs, and recovery for a fictional system that can act through software.
try · 18 min · inai-originalDecompose work into tasks, relationships, judgment, and responsibility before making whole-job predictions.
try · 18 min · inai-originalPlace review where an AI-assisted workflow becomes consequential, uncertain, sensitive, or difficult to reverse.
try · 16 min · inai-originalEvaluate origin, context, corroboration, provenance, and consequences without pretending that appearance alone detects truth.
try · 18 min · inai-originalBuild a sensitive-context checklist before giving text, files, images, or records to an AI system.
try · build · 8 min · inai-originalTurn a broad product promise into a claim-and-evidence comparison sheet.
try · build · 10 min · inai-originalClassify a system from observable behavior rather than marketing language.
try · build · 8 min · inai-originalProduce an access, action, and reversibility checklist before an agent can affect software or people.
try · build · 10 min · inai-originalTurn “a human is involved” into a workflow approval map with timing, evidence, authority, and recovery.
try · build · 10 min · inai-originalCreate a task and responsibility inventory without turning one capability into a whole-job forecast.
try · build · 12 min · inai-originalCreate an adaptable internal rules outline grounded in real tasks, data, review, ownership, and learning.
try · build · 15 min · inai-originalCreate a study-method guide that uses AI for feedback and explanation while preserving retrieval, effort, evidence, and authorship.
try · build · 10 min · inai-originalBuild an evidence and corroboration checklist for image, audio, video, or screenshots.
try · build · 10 min · inai-originalCreate a purpose, data, capability, cost, and risk table for one real use.
try · build · 12 min · inai-originalCreate a context, constraint, evidence, and output-format plan without treating prompting as magic.
try · build · 8 min · inai-originalBuild a five-sentence explain-back that separates AI, models, products, current limits, and human responsibility.
try · build · 8 min · inai-originalCreate a date, version, source, and relevance worksheet for a changing AI claim.
try · build · 10 min · inai-originalCreate a shared expectations template for learning, privacy, authorship, fairness, help, and revision.
try · build · 15 min · inai-originalBuild a value, risk, reversibility, evidence, and ownership decision tree before changing a workflow.
try · build · 12 min · inai-originalReference and context
Definitions, profiles, milestones, collections, and registries.A source and evidence shelf showing support type, scope, review timing, uncertainty, and correction status.
examine · 10 min · inai-curationArtificial intelligence is a field and family of systems that infer how to produce outputs from inputs in pursuit of explicit or implicit objectives.
orient · 2 min · inai-explanationAn algorithm is a defined procedure for turning inputs into results through a sequence of operations or rules.
orient · 2 min · inai-explanationMachine learning is an approach in which a model’s behavior is shaped by finding patterns in data or experience rather than only by fixed task rules.
orient · 2 min · inai-explanationDeep learning is machine learning built with neural networks containing multiple layers of learned transformations.
orient · 2 min · inai-explanationA neural network is a parameterized computing structure that passes information through connected units and transformations to learn an input–output relationship.
orient · 2 min · inai-explanationA model is a learned or designed representation that maps inputs to outputs for a defined task or family of tasks.
orient · 2 min · inai-explanationParameters are internal numerical values that a model’s training procedure adjusts to shape its behavior.
orient · 2 min · inai-explanationTraining is the process of adjusting a model to improve performance on an objective using data, feedback, or interaction.
orient · 2 min · inai-explanationInference is running a trained model on an input to produce a prediction, generation, score, or other output.
orient · 2 min · inai-explanationA dataset is an organized collection of examples or records used for training, evaluation, analysis, or system operation.
orient · 2 min · inai-explanationA label is a target value or category attached to an example for training or evaluating a supervised model.
orient · 2 min · inai-explanationA feature is an input property or representation used by a model when producing an output.
orient · 2 min · inai-explanationA prediction is a model’s estimated output for a given input, including a class, value, probability, sequence, or next element.
orient · 2 min · inai-explanationClassification is assigning an input to one or more defined categories, often with a score or probability for each.
orient · 2 min · inai-explanationRegression is a modeling task that predicts a numerical value rather than a discrete category.
orient · 2 min · inai-explanationGeneralization is a model’s ability to perform usefully on relevant examples or conditions not seen during training.
orient · 2 min · inai-explanationEvaluation is the structured measurement and investigation of how a model or system performs against defined questions, criteria, and contexts.
orient · 2 min · inai-explanationA benchmark is a standardized task, dataset, procedure, and metric used to compare model or system performance.
orient · 2 min · inai-explanationOverfitting occurs when a model fits training examples or their noise too closely and performs less well on relevant new cases.
orient · 2 min · inai-explanationDistribution shift is a meaningful difference between the data or conditions used to develop a model and those encountered later.
orient · 2 min · inai-explanationGenerative AI produces new text, images, audio, video, code, or other content in response to context and controls.
orient · 2 min · inai-explanationA foundation model is trained broadly enough to be adapted or used across multiple downstream tasks and applications.
orient · 2 min · inai-explanationA large language model, or LLM, is a parameter-rich model trained on language sequences to predict and generate tokens in context.
orient · 2 min · inai-explanationA token is a unit into which a model’s input or output is divided for processing.
orient · 2 min · inai-explanationAn embedding is a numerical vector that represents an item so useful relationships can be measured or learned.
orient · 2 min · inai-explanationA context window is the limited amount of tokenized information a model can consider together during one generation or inference.
orient · 2 min · inai-explanationA prompt is input supplied to a generative model or system to shape the task, context, constraints, or desired output.
orient · 2 min · inai-explanationA system instruction is configuration context intended to guide a model’s behavior across a session or task, often with priority over ordinary user input.
orient · 2 min · inai-explanationTemperature is a generation setting that changes how strongly token sampling favors higher-probability options.
orient · 2 min · inai-explanationA multimodal model processes or produces more than one kind of data, such as text, images, audio, video, or sensor signals.
orient · 2 min · inai-explanationA hallucination is generated content that is unsupported, inaccurate, or invented while often remaining fluent and plausible.
orient · 2 min · inai-explanationGrounding connects a model’s output to selected external information, observations, rules, or evidence relevant to the task.
orient · 2 min · inai-explanationRetrieval-augmented generation, or RAG, retrieves relevant material and places it in context for a generative model to use.
orient · 2 min · inai-explanationFine-tuning is additional training that adapts a pretrained model to selected tasks, domains, examples, or behavior goals.
orient · 2 min · inai-explanationAlignment is the effort to make an AI system’s behavior accord with specified human intentions, values, rules, or objectives.
orient · 2 min · inai-explanationA model card is structured documentation describing a model’s intended uses, evaluation, limitations, development context, and relevant risks.
orient · 2 min · inai-explanationAn AI agent is a system that continues across steps toward a goal and can select or use actions, tools, or information within defined controls.
orient · 2 min · inai-explanationAn agentic system is the full technical and organizational arrangement that enables goal-directed continuation and action.
orient · 2 min · inai-explanationTool use is a system’s ability to invoke an external function, service, interface, or device and incorporate the result.
orient · 2 min · inai-explanationFunction calling lets a model propose a structured function name and arguments for software to validate and optionally execute.
orient · 2 min · inai-explanationA workflow is an organized sequence of tasks, decisions, handoffs, information, and checks used to produce an outcome.
orient · 2 min · inai-explanationAutomation is the use of technology to perform part or all of a process with reduced direct human execution.
orient · 2 min · inai-explanationOrchestration coordinates multiple tasks, models, tools, services, agents, or people so their outputs form one process.
orient · 2 min · inai-explanationMemory is information a system retains or can retrieve for use beyond the immediate step or context.
orient · 2 min · inai-explanationState is the information that describes a system or task at a particular point and is carried into later steps.
orient · 2 min · inai-explanationA permission is an authorized ability to access information, use a tool, or perform an action under defined conditions.
orient · 2 min · inai-explanationHuman in the loop describes a process where a person performs a defined review, decision, correction, or approval within a system’s operating cycle.
orient · 2 min · inai-explanationHuman oversight is the wider capacity of responsible people and institutions to understand, govern, monitor, challenge, and intervene in an AI system.
orient · 2 min · inai-explanationA guardrail is a preventive, detective, or corrective control intended to keep system behavior within defined boundaries.
orient · 2 min · inai-explanationRecovery is the ability to stop, contain, reverse, retry, compensate for, or safely continue after failure or interruption.
orient · 2 min · inai-explanationObservability is the capacity to understand a system’s relevant internal state and behavior from designed records, signals, and traces.
orient · 2 min · inai-explanationAutonomy is the degree of freedom a system has to choose, continue, or act without a person directing each step.
orient · 2 min · inai-explanationBias is a systematic tendency in data, measurement, design, model behavior, or institutions that shifts results in particular directions.
orient · 2 min · inai-explanationFairness is the context-dependent quality of treating people or groups according to justified criteria, rights, needs, and procedures.
orient · 2 min · inai-explanationExplainability is the extent to which relevant people can receive a useful account of how or why a system produced a result.
orient · 2 min · inai-explanationTransparency is making relevant information about a system, its use, evidence, limits, and governance available to the people who need it.
orient · 2 min · inai-explanationProvenance is the recorded origin, custody, transformation, and history of data, media, evidence, or a system artifact.
orient · 2 min · inai-explanationSynthetic media is text, imagery, audio, video, or other media created or materially altered by computational generation.
orient · 2 min · inai-explanationA deepfake is highly realistic synthetic or manipulated media that depicts a person, voice, event, or action in a materially false way.
orient · 2 min · inai-explanationMisinformation is false, inaccurate, or materially misleading information shared regardless of whether the sharer intended to deceive.
orient · 2 min · inai-explanationPrivacy concerns appropriate control, expectations, and risk around how information about people is collected, inferred, used, shared, retained, and deleted.
orient · 2 min · inai-explanationPersonal data is information that relates to an identified or reasonably identifiable person, directly or through combination and inference.
orient · 2 min · inai-explanationSensitive data is information whose exposure, misuse, inference, or alteration could create substantial harm or requires heightened protection.
orient · 2 min · inai-explanationSecurity protects systems and information against unauthorized access, change, disclosure, disruption, or destruction.
orient · 2 min · inai-explanationAn AI incident is an event where an AI system causes, contributes to, or creates a credible risk of harm, failure, or rights impact.
orient · 2 min · inai-explanationA high-risk use is an application where failure, misuse, or unequal performance can create serious consequences for people, rights, safety, livelihood, or essential services.
orient · 2 min · inai-explanationVerification is the process of checking whether a claim, output, action, identity, or result is adequately supported for its intended use.
orient · 2 min · inai-explanationA primary source is original evidence or a first-hand record produced by the people, instrument, institution, or process directly involved.
orient · 2 min · inai-explanationEvidence status is a visible label describing how strongly and consistently current sources support a particular claim.
orient · 2 min · inai-explanationAI literacy is the context-sensitive knowledge, skills, and judgment needed to understand, use, evaluate, and shape AI systems responsibly.
orient · 2 min · inai-explanationCompute is the processing capacity and work supplied by hardware to train, run, and operate digital models and systems.
orient · 2 min · inai-explanationAn accelerator is specialized hardware designed to perform selected computations faster or more efficiently than a general-purpose processor.
orient · 2 min · inai-explanationA data center is a facility and operating environment that houses computing, storage, networking, power, cooling, and security systems.
orient · 2 min · inai-explanationOpen source describes software whose source code is available under a license granting defined rights to use, inspect, modify, and redistribute it.
orient · 2 min · inai-explanationAn open model is a model for which selected artifacts and usage rights are made publicly available, with the degree of openness stated explicitly.
orient · 2 min · inai-explanationA closed model is a model whose key artifacts, internal details, or operating access are not publicly released beyond provider-defined interfaces and terms.
orient · 2 min · inai-explanationAn API, or application programming interface, is a defined way for software components to request functions or exchange data.
orient · 2 min · inai-explanationRobotics is the field of designing, building, sensing, controlling, and operating machines that act in the physical world.
orient · 2 min · inai-explanationAGI, or artificial general intelligence, is a contested family of terms for AI with broad, flexible capability across many tasks or environments.
orient · 2 min · inai-explanationASI, or artificial superintelligence, is a speculative term for AI capability that substantially exceeds humans across most or all relevant cognitive domains.
orient · 2 min · inai-explanationBuild a usable map of the field before choosing a deeper direction.
orient · understand · try · build · 60 min · inai-explanationReplace headline fear with a task-level map of change, judgment and uncertainty.
orient · understand · try · build · 60 min · inai-explanationUse assistance without outsourcing your identity or exposing unnecessary personal data.
orient · understand · try · build · 60 min · inai-explanationMove beyond both panic and uncritical enthusiasm.
orient · understand · try · build · 60 min · inai-explanationBuild an assessment and classroom-use framework, not only a policy opinion.
orient · understand · try · build · 60 min · inai-explanationUse assistance while keeping the knowledge and judgment you need.
orient · understand · try · build · 60 min · inai-explanationIdentify one plausible use, one risk boundary and one next step without a sales funnel.
orient · understand · try · build · 60 min · inai-explanationConnect literacy, workflow redesign, permissions and change management.
orient · understand · try · build · 60 min · inai-explanationFrame deployment around affected people, evidence, oversight and recourse.
orient · understand · try · build · 60 min · inai-explanationConnect models, tools, state, evaluation, permissions and recovery.
orient · understand · try · build · 60 min · inai-explanationInspect uncertainty, incentives and disagreement instead of being asked to agree.
orient · understand · try · build · 60 min · inai-explanationTuring helped establish what computation can mean mathematically and later asked how claims about machine intelligence might be made testable.
understand · examine · 8 min · inai-explanationMcCarthy helped give AI an institutional name and advanced programming and logical formalisms for representing and reasoning with knowledge.
understand · examine · 8 min · inai-explanationMinsky helped establish major AI institutions and explored intelligence as an interaction among many specialized processes rather than one simple mechanism.
understand · examine · 8 min · inai-explanationHinton sustained and expanded neural-network research through periods of skepticism, helping develop methods and research communities central to modern deep learning.
understand · examine · 8 min · inai-explanationBengio helped build the conceptual foundations of deep learning and now concentrates much of his research and public work on advanced-AI safety.
understand · examine · 8 min · inai-explanationLeCun helped turn convolutional neural networks into practical systems for visual and document recognition and now pursues architectures intended to learn predictive models of the world.
understand · examine · 8 min · inai-explanationLi led the creation of ImageNet with collaborators, helping make large, shared datasets and benchmarks central to progress in computer vision.
understand · examine · 8 min · inai-explanationPearl developed influential graphical and mathematical tools for reasoning under uncertainty and for making causal assumptions explicit.
understand · examine · 8 min · inai-explanationBreazeal pioneered social robotics and human–robot interaction by studying how machines can use expressive cues and participate in structured social exchanges.
understand · examine · 8 min · inai-explanationGebru's research connects technical evidence about datasets and model performance with questions about who builds AI, who is affected and which institutions control research.
understand · examine · 8 min · inai-explanationBuolamwini combined empirical auditing, public communication and advocacy to make disparities in automated facial analysis visible and actionable.
understand · examine · 8 min · inai-explanationHassabis co-founded and leads DeepMind, an institution that has combined machine learning, search and large research teams in systems such as AlphaGo and AlphaFold.
understand · examine · 8 min · inai-explanationTwenty-one sourced milestones show what changed, what did not, and how people, concepts, institutions, and governance connect.
orient · understand · examine · 45 min · inai-curationBetween 1943 and 1948, electronic computation, simplified mathematical neurons and feedback-oriented cybernetics created several foundations later absorbed into AI.
understand · examine · 3 min · inai-explanationAlan Turing’s paper replaced the poorly specified question “Can machines think?” with an imitation-game setup and examined learning machines, objections and limits.
understand · examine · 3 min · inai-explanationThe Dartmouth Summer Research Project brought researchers together around the proposal that aspects of learning and intelligence might be described precisely enough for machines to simulate.
understand · examine · 3 min · inai-explanationFrank Rosenblatt’s perceptron work demonstrated a trainable classifier whose weights could change after errors, giving early neural-network research a concrete algorithm and machine project.
understand · examine · 3 min · inai-explanationProjects such as DENDRAL and MYCIN encoded domain knowledge, rules and inference procedures to support scientific or medical reasoning in bounded settings.
understand · examine · 3 min · inai-explanationAfter ambitious predictions met technical limitations and skeptical reviews, some AI funding and institutional support contracted in the UK and United States.
understand · examine · 3 min · inai-explanationRumelhart, Hinton and Williams demonstrated an effective procedure for adjusting weights through multiple layers so hidden units could learn useful internal features.
understand · examine · 3 min · inai-explanationCosts of maintaining brittle knowledge bases, specialized hardware shifts and unmet commercial expectations contributed to a downturn in expert-system businesses and AI investment.
understand · examine · 3 min · inai-explanationAcross the 1990s, probabilistic models, support-vector machines, decision trees, ensembles and benchmark-based evaluation made data-driven prediction a larger part of AI research.
understand · examine · 3 min · inai-explanationImageNet organized millions of web images into a large hierarchy and later supported a common large-scale recognition challenge.
understand · examine · 3 min · inai-explanationKrizhevsky, Sutskever and Hinton’s ImageNet system combined convolutional networks, large labeled data, GPUs and training techniques to improve image-classification results sharply.
understand · examine · 3 min · inai-explanationEfficient word embeddings made linguistic patterns usable as vectors, while sequence-to-sequence neural models learned mappings between variable-length text sequences.
understand · examine · 3 min · inai-explanationThe transformer architecture used self-attention and feed-forward layers instead of recurrent or convolutional sequence processing in its core translation design.
understand · examine · 3 min · inai-explanationOECD members adopted principles connecting trustworthy AI with human rights, democratic values, transparency, robustness, accountability and inclusive benefit.
understand · examine · 3 min · inai-explanationGPT-3 demonstrated that a scaled autoregressive language model could perform many tasks from instructions or examples in its input without task-specific weight updates.
understand · examine · 3 min · inai-explanationUNESCO’s General Conference adopted a normative recommendation covering human rights, inclusion, environmental effects, data governance, education, culture and other policy areas.
understand · examine · 3 min · inai-explanationChatGPT’s research preview put a capable instruction-following language model behind a simple dialogue interface that people could try without building software.
understand · examine · 3 min · inai-explanationAI RMF 1.0 organized risk work around governing, mapping, measuring and managing AI systems across their lifecycle.
understand · examine · 3 min · inai-explanationSystems announced during 2023 connected language-model interfaces with image input and, across model families, wider combinations of text, image, audio or video.
understand · examine · 3 min · inai-explanationResearch patterns such as ReAct, product function-calling interfaces and emerging connection protocols made it easier to combine a model with retrieval, code, APIs and multi-step action loops.
understand · examine · 3 min · inai-explanationRegulation (EU) 2024/1689 entered into force on 1 August 2024, establishing a binding, risk-based legal framework with rules for prohibited practices, high-risk systems, transparency and general-purpose models.
understand · examine · 3 min · inai-explanationExternal resources
Courses, reports, tools, and videos annotated by inAi.A free, structured introduction that moves from definitions and problem solving through machine learning, neural networks and social implications.
examine · 1800 min · external-courseA concise non-technical course on what AI can and cannot do, how AI projects are organized, and how organizations can adopt it responsibly.
examine · 414 min · external-courseA six-hour Level 1 OpenLearn introduction linking the history and mechanisms of AI with ethical and social questions.
examine · 360 min · external-courseA short beginner learning path that introduces AI concepts through the Azure ecosystem.
examine · 120 min · external-publicationA modular technical course covering regression, classification, data, neural networks, embeddings, large language models, production systems and fairness.
examine · 900 min · external-courseA short, notebook-centered introduction to building and validating simple machine-learning models.
examine · 240 min · external-courseA free nine-lesson course that teaches deep learning by building working applications with modern libraries.
examine · 810 min · external-courseA demanding seven-week Python course covering foundational AI algorithms through lectures and programming projects.
examine · 4200 min · external-courseA changing catalogue of courses on LLMs, agents, robotics, reinforcement learning, computer vision, audio, diffusion and related open-model tooling.
examine · 60 min · external-courseA living technical course on building and evaluating agent workflows with current open-source frameworks.
examine · 960 min · external-courseA browser visualization for experimenting with small neural networks and watching decision boundaries change.
examine · 45 min · external-publicationA browser tool for training simple image, sound or pose classifiers from examples without writing code.
examine · 45 min · external-publicationA visual video collection developing intuition for neural networks, gradient descent and related mathematics.
examine · 120 min · external-publicationA widely used illustrated walkthrough of the original transformer architecture and attention flow.
examine · 45 min · external-publicationA collection of visual essays explaining selected machine-learning methods and evaluation ideas.
examine · 60 min · external-publicationA facilitated classroom unit in which students debate AI-related issues from different stakeholder positions.
examine · 150 min · external-publicationAn international framework for designing student AI competencies around human-centred values, ethics, foundations, applications and system design.
examine · 180 min · external-publicationAn international reference for teacher competencies in AI, spanning human-centred mindset, ethics, foundations, pedagogy and professional learning.
examine · 180 min · external-publicationA K–12 guidance initiative organizing AI education around perception, representation and reasoning, learning, natural interaction and societal impact.
examine · 120 min · external-publicationA 2025 toolkit for education authorities and schools developing AI guidance, policies and stakeholder processes.
examine · 180 min · external-publicationA comparative portal for AI policies, indicators, incidents, tools and issue-specific analysis.
examine · 90 min · external-publicationNIST’s voluntary framework for organizing how AI risks are governed, mapped, measured and managed.
examine · 240 min · external-publicationA companion profile that identifies generative-AI risks and suggested actions within the NIST AI RMF structure.
examine · 300 min · external-publicationThe 2026 edition of Stanford HAI’s annual synthesis of evidence on AI research, industry, policy, public opinion and societal effects.
examine · 480 min · external-publicationAn official European Commission Q&A explaining the AI Act’s AI-literacy provision and practical implementation considerations.
examine · 45 min · external-publicationAn IEA report examining the relationship between AI, data-centre electricity demand, energy-system applications and policy choices.
examine · 240 min · external-publicationA July 2026 ILO brief examining generative-AI exposure, observed labour-market disruption and preparedness across ASEAN.
examine · 90 min · external-publicationThe current C2PA technical specification for attaching and validating cryptographically signed provenance information about digital content.
examine · 360 min · external-publicationNIST’s voluntary framework for identifying and managing privacy risk through organizational functions and profiles.
examine · 240 min · external-publicationW3C WAI guidance on planning captions, transcripts, audio descriptions and accessible media players.
examine · 60 min · external-publicationA brisk route from early chess programs and symbolic rules to neural networks, transformers, benchmarks, and scaling laws. Use it to see that today’s wave has a long technical lineage—not to treat every forecast near the end as settled.
orient · understand · 13 min · external-videoA visual construction of a small digit-recognition network: pixels become activations, layers transform them, and weights and biases determine the result. The video motivates what ‘learning’ changes, while deliberately leaving the training algorithm to the next chapter.
orient · understand · 19 min · external-videoA visual tour of how a GPT-style transformer turns tokens into vectors, updates them through attention and feed-forward blocks, then produces a probability distribution for the next token. It explains repeated prediction and sampling without pretending that next-token generation is simple lookup.
orient · understand · 28 min · external-videoAn approachable 2023 explanation of fluent but unsupported model outputs, with examples of contradiction, fabrication, and irrelevance. Its prompting advice is useful as risk reduction, but the correct habit is still to verify consequential claims against reliable sources.
orient · understand · 10 min · external-videoA clear practical ladder from a prompted chatbot, to a workflow with human-defined control logic, to an agent loop that chooses tools and iterates toward a goal. Keep the ladder as a useful comparison—not as the only accepted definition of an agent.
orient · understand · 11 min · external-videoILO researchers explain why exposure is estimated from the tasks inside occupations rather than from dramatic job headlines. The 2023 study models potential automation and augmentation, then emphasizes job quality, gender, access, worker voice, redeployment, and the gap between technical possibility and adoption.
orient · understand · 57 min · external-videoA compact introduction to media provenance: verified capture, edit histories, and chains from creation to publication can add evidence about origin. WITNESS also shows why an authenticity signal must not become a truth badge or exclude people who need ordinary devices, anonymity, or safe editing.
orient · understand · 4 min · external-videoJoy Buolamwini uses repeated failures of facial-analysis systems to make algorithmic bias concrete, then traces harm from training data and development teams into policing, credit, hiring, and other institutions. The talk is historically important; its statistics and regulatory descriptions remain dated to 2016.
orient · understand · 9 min · external-videoA four-minute account of how servers and cooling drive data-centre electricity demand, why demand is geographically concentrated, and how the IEA’s 2025 Base Case reaches about 945 TWh in 2030. The video gives a system-scale forecast, not a universal per-prompt energy number.
orient · understand · 5 min · external-videoAn extended conversation about competing meanings of AGI, singularity, and superintelligence; limitations attributed to current transformer systems; and predictive-coding, neurosymbolic, evolutionary, memory, and world-model alternatives. Its value is the disagreement and architecture map—not the guest’s confident 2029 forecast.
orient · understand · 64 min · external-video