liveBookmanning.com

What you will learn in Inside AI

Over 150 billion purchases per year use this author’s AI

by Akli Adjaoute

Read it on liveBook Get the book at manning.com

Chapter by chapter

Concepts explained in this book

• artificial intelligenceArtificial intelligence is the field of computational methods that perform tasks associated with human intelligence, such as learning, reasoning, perception, and decision-making.
• language modelA large language model is a machine-learning model trained on large text corpora to process and generate human-language text.
• deep learningA deep neural network is a neural network containing multiple hidden layers that learn hierarchical representations from simpler features.
• multi-step reasoningLogical reasoning is the process of deriving conclusions from information by applying rules of consistency and inference.
• contextContext is the surrounding situation, knowledge, and circumstances that influence the interpretation of an utterance or event.
• machine learningMachine learning is a computational approach in which models learn patterns from data to perform tasks without being explicitly programmed for every case.
• machine translationAutomatic translation is the computational conversion of content from one human language to another.
• clinical reasoningClinical reasoning is the process of interpreting clinical information and applying medical knowledge to form and revise diagnostic judgments and select appropriate patient-care actions.
• understandingUnderstanding is the capacity to form meaningful and context-sensitive representations of information, situations, or experiences.
• automationAutomation is the execution of tasks or processes by systems with reduced direct human intervention.
• AI agentAn autonomous system is a system that performs functions and makes operational decisions with limited direct human control.
• misinformationMisinformation is false or misleading information disseminated without the defining intention to manipulate or deceive.
• ChatGPTChatGPT is a conversational artificial-intelligence system that generates responses to natural-language prompts.
• generative modelingGenerative artificial intelligence is an approach in which models produce new content such as text, images, audio, or code from learned patterns.
• pattern recognitionPattern recognition is the identification of recurring regularities in data without necessarily representing their conceptual or real-world meaning.
• imaginationImagination is the human mental faculty for forming and manipulating ideas, scenes, possibilities, and entities that are not directly present.
• expert systemAn expert system is a computer-based system that separates domain-specific knowledge from procedural methods to mimic human expert decision-making in a specific domain.
• domain knowledgeDomain knowledge is specialized understanding of the facts, practices, constraints, and terminology of a particular subject area.
• reliabilityReliability is the degree to which a system performs its intended functions consistently and correctly under specified conditions.
• creativityCreativity is the human cognitive capacity to generate novel ideas, products, or solutions by forming and transforming connections.
• risk ratingA risk score is a numerical estimate of the likelihood or severity of an undesirable event such as fraudulent activity.
• intelligenceIntelligence is the cognitive capacity to make decisions, solve problems, and respond appropriately to situations.
• unsupervised learningUnsupervised learning is a machine-learning technique in which an algorithm analyzes unlabeled data to discover patterns or groupings.
• technological singularityThe technological Singularity is a hypothetical point at which rapidly advancing artificial intelligence produces transformative and potentially uncontrollable changes beyond human ability to keep up.
• parallel inferenceParallel inference is the simultaneous execution of logical or knowledge-based reasoning operations across multiple processing units.
• data miningData mining is the computational process of discovering patterns, relationships, and useful insights in large datasets.
• brainThe human brain is the central organ of the human nervous system, coordinating cognition, perception, behavior, and bodily functions.
• fraud preventionFraud prevention comprises methods for identifying and stopping deceptive transactions or activities before they cause loss.
• medical diagnosisMedical diagnosis is the identification of a disease or medical condition based on symptoms, observations, and clinical evidence.
• adaptive aiLearning and adaptation is the capability of a system to acquire improved behavior or modify its responses based on experience or changing conditions.
• job lossThe reduction or disappearance of employment opportunities for workers or occupational roles.
• causationCausality is the relationship in which one event, condition, or variable produces or influences another.
• fitnessA fitness function is a scoring function that evaluates the quality of a candidate solution in an optimization process.
• backpropagationBackpropagation is a neural-network training technique that computes weight adjustments by propagating output error backward through the network.
• hidden layerA hidden layer is an intermediate neural-network layer that transforms representations between the input and output layers.
• goalAn agent goal is a desired state or condition that an agent seeks to achieve.
• genetic algorithmA genetic algorithm is an optimization technique that evolves a population of candidate solutions through evaluation, selection, recombination, and random variation.
• intuitionIntuition is the capacity to reach judgments or insights through immediate, non-deliberative perception rather than explicit step-by-step analysis.
• hallucinationA hallucination is AI-generated content that is nonsensical, unsupported, or false despite appearing to be a meaningful response.
• computer visionComputer vision is the field concerned with enabling computers to interpret and analyze visual images or video.
• labeled dataLabeled data is a dataset whose examples are paired with target labels or outcomes.
• fabricated contentFabricated content is media or information that has been artificially created or altered to appear authentic despite not representing genuine events or facts.
• inference engineAn inference engine is a program that evaluates known facts against rules to derive or modify facts and control an expert system's reasoning loop.
• robotA robot is a programmable machine capable of sensing, processing information, and carrying out physical actions.
• roboticsRobotics is the field concerned with designing, building, and controlling robots and robotic systems.
• deep fakeA deepfake is digitally generated or altered media that realistically depicts a person or event in a way that is not authentic.
• fuzzy logicFuzzy logic is a mathematical logic framework that represents and reasons with degrees of truth rather than only binary true-or-false values.
• personalizationPersonalization is the adaptation of a system’s behavior or outputs to the characteristics, preferences, or context of an individual user.
• decision treeA decision tree is a hierarchical predictive model that recursively splits data according to attribute-based questions to reach decisions or predictions.
• neuronA neuron is a nerve cell that transmits information through electrical impulses and chemical signals.
• disinformationDisinformation is false or misleading information intentionally spread to manipulate public perception or achieve an agenda.
• interpretabilityInterpretability is the degree to which a model’s decisions or outputs can be understood and explained by people.
• normalizationNormalization is the transformation of numeric values to a specified bounded range so that they can be processed consistently by a model.
• crossoverCrossover is a genetic-algorithm operation that combines parts of two candidate solutions to create a new candidate.
• mutationMutation is a genetic-algorithm operation that introduces a random modification into a candidate solution.
• populationA population is a collection of candidate solutions maintained by a genetic algorithm.
• objective functionAn objective function is a function that assigns a value representing the performance or desirability of a candidate solution.
• selectionSelection is a genetic-algorithm operation that retains candidate solutions with comparatively high fitness for further evolution.
• Traveling Salesman problemThe traveling salesman problem is an optimization problem that seeks the best route visiting each location exactly once and returning to the starting location.
• AGORAAGORA is a proprietary programming language designed to represent and execute the smart agent paradigm.
• stabilizeStabilize is an AGORA keyword that expresses an acceptable or desirable state for an agent.
• unstabilizeUnstabilize is an AGORA keyword that expresses a forbidden or unacceptable state for an agent.
• entity profileAn entity profile is a structured representation of information and behavioral history associated with a particular entity.
• input layerAn input layer is the neural-network layer that receives the input data.
• output layerAn output layer is the neural-network layer that produces the model’s final prediction or output.
• cyberbullyingThe use of digital communication or online platforms to harass, intimidate, or harm another person.
• intelligence explosionAn intelligence explosion is a hypothesized rapid increase in an artificial system’s intellectual capabilities driven by its ability to improve the design of increasingly capable systems.
• recursive self-improvementRecursive self-improvement is the process by which an intelligent system repeatedly improves its own design or capabilities, enabling further improvements.
• superintelligenceAn ultra-intelligent machine is a hypothetical machine whose intellectual abilities greatly exceed those of any human.
• case-based reasoningCase-based reasoning is an artificial-intelligence technique that solves new problems by retrieving, adapting, and revising solutions from similar past cases.
• transparencyTransparency is the property of making a system's data, processes, or decision basis sufficiently visible for scrutiny.
• pixelA pixel is the smallest addressable picture element in a digital image, represented here by color-component values.
• curiosityCuriosity is the motivation to seek information, explanations, or experiences beyond what is already known.
• cognitionLiving intelligence is the adaptive information-processing capacity exhibited by biological organisms and their cells.
• common senseCommon-sense knowledge is informal background knowledge about ordinary situations, entities, and expectations that people typically apply without explicit instruction.
• black boxA black-box algorithm is an algorithm whose internal operations and reasoning are not readily interpretable from its inputs and outputs.
• synapseA synapse is a specialized junction through which a neuron communicates with another neuron or target cell.
• fairnessFairness is the property of treating people equitably and avoiding unjustified disparities in system outcomes.
• risk assessmentRisk assessment is the process of evaluating the likelihood and potential impact of uncertain adverse events.
• synthetic mediaSynthetic media is audio, video, imagery, or other media generated or substantially altered by computational methods.
• clusteringClustering is the computational task of grouping data items according to similarities without requiring predefined categories.
• alpha-beta pruningAlpha-beta pruning is a search optimization algorithm that eliminates game-tree branches that cannot affect the selected minimax evaluation.
• game treeA game tree is a branching representation whose nodes denote game states and whose edges denote states reachable through legal moves.
• heuristicA heuristic is a practical rule or approximation for making decisions when exhaustive computation is impractical.
• GPT modelGPT-3 is a generative language model developed by OpenAI for processing and generating human-language text.
• fuzzy setA fuzzy set is a set in which elements have graded rather than strictly binary membership.
• if-then ruleAn if-then rule is a conditional knowledge representation that specifies an action or conclusion when stated conditions are satisfied.
• safe deploymentSafe deployment is the controlled release and operation of a system with measures to reduce harmful failures and unintended consequences.
• brain-computer interfaceA brain-computer interface is a system that records neural activity and translates it into signals for controlling or communicating with an external computer or device.
• human-machine intelligence mergerA human-machine hybrid is a hypothetical entity combining human biological functions with integrated machine components.
• mind uploadingMind uploading is the hypothetical transfer of a person's mental states or cognitive processes into a computational system.
• Moore's lawMoore's Law is the empirical observation that the number of transistors on an integrated circuit tends to increase exponentially, historically doubling about every two years.
• conversational aiA virtual assistant is a software system that performs tasks or provides information through interaction with users, often using natural-language requests.
• adversarial attackAn adversarial example is an input deliberately altered to cause a machine-learning model to produce an incorrect or unreliable output.
• reinforcement learningReinforcement learning is a machine-learning technique in which an agent learns behavior by interacting with an environment and receiving feedback for its actions.
• trustTrustworthiness is the degree to which a system can be relied upon to behave safely, reliably, transparently, and appropriately.
• business ruleA business rule is a declarative condition or constraint that specifies an organizational decision or action for particular circumstances.
• business rules management systemA business rules management system is software for authoring, managing, governing, and executing business rules separately from application code.
• Bayesian networkA Bayesian network is a probabilistic model represented by a directed acyclic graph whose nodes are variables and whose directed edges encode conditional dependencies.
• knowledge baseA knowledge base is a repository of domain-specific facts and rules that represents accumulated expert knowledge for an intelligent system.
• working memoryWorking memory is the component of an expert system that stores input data and intermediate facts produced during reasoning.
• ai projectAn AI project is an organized effort to design, develop, and deploy an artificial-intelligence solution for a defined objective.
• incoherence detectionIncoherence detection is the identification of objects or events that violate expected contextual or relational consistency.
• false positiveA false positive is an alert or classification indicating a condition that is not actually present.
• rule-authoring environmentA rule-authoring environment is a software interface for creating, modifying, and testing business rules.
• rule repositoryA rule repository is an organized storage component in which business rules are defined, managed, and retrieved.
• tokenA token is a sequence of characters, word fragment, or related text unit processed by a language model.
• securitySecurity is the protection of systems, data, and operations from threats, misuse, disruption, or unauthorized access.
• mailboxA mailbox is a message-storage and communication channel through which an agent receives messages from other agents or external sources.
• symbolic AISymbolic AI is an artificial-intelligence approach that represents knowledge explicitly with symbols, rules, and logic.
• semantic gapA semantic gap is the mismatch between data representations that a computational system can process and the deeper real-world meanings they signify.
• symbolic reasoningSymbolic reasoning is reasoning that derives conclusions by manipulating explicit symbols according to formal rules and logic.
• convolutional neural networkA convolutional neural network is a neural-network model that applies learned convolutional filters to structured data such as images.
• RGBThe RGB color system represents colors as combinations of red, green, and blue component values.
• Vision TransformerA vision transformer is a transformer-based model that processes image representations for computer-vision tasks.
• bayesian statisticsBayes' theorem is a probability formula for updating the probability of a hypothesis using new evidence and prior probabilities.
• conditional probabilityConditional probability is the probability of an event given that another event or condition is known to have occurred.
• directed acyclic graphA directed acyclic graph is a directed graph containing no path that returns to its starting node.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.