AI comes with a lot of new words. Knowing the terms helps you see the full picture. Browse, learn, and build your own glossary.
200 / 2551 terms
A type of language model where the internal parameters are constrained to values of only -1 or 1, drastically reducing memory usage compared to standard 16-bit or 8-bit precision models.
The sixth generation of 10-nanometer class manufacturing technology, which allows for denser, faster, and more power-efficient memory chips.
A health guideline recommended to reduce digital eye strain by looking at something 20 feet away for 20 seconds every 20 minutes.
An HTTP status code indicating a temporary redirect, where the requested URL has been moved to a different location for the time being.
A federal program that requires pharmaceutical manufacturers to provide outpatient drugs to eligible healthcare organizations and covered entities at significantly reduced prices.
The ability of a computational system to perceive, reason about, and interact with the three-dimensional physical environment and its geometric properties.
A technology that stacks additional L3 memory vertically on top of or beneath the processor to increase data access speeds.
An educational technology company founded in 2003 that develops online learning platforms for mathematics and literacy.
The process of creating digital content that accounts for three spatial dimensions plus the temporal dimension of time.
A technology that captures and reconstructs both the 3D shape and the temporal movement of objects over time.
Regulatory requirement by the U.S. Food and Drug Administration for medical device manufacturers to prove a product is safe and effective by comparison to a legally marketed device.
A venture capital firm based in Austin, Texas, specializing in early-stage investments across technology and life sciences.
Singapore's lead public sector agency that spearheads economic-oriented research to advance scientific discovery and develop innovative technology.
Andreessen Horowitz, a prominent Silicon Valley venture capital firm that invests in technology companies.
Short for Agent-to-Agent, a protocol where software entities communicate to negotiate and align decisions rather than just exchanging raw data.
A framework where multiple autonomous software agents interact and collaborate to perform complex tasks without direct human intervention.
The 64-bit instruction set architecture for ARM processors, commonly found in mobile devices and modern desktop computers.
An advanced search algorithm that enables multiple AI models to collaborate and optimize decision-making processes during inference.
A global research center based at MIT that conducts randomized impact evaluations to reduce poverty through evidence-based policy.
The process of removing or disabling specific components of an artificial intelligence system to understand their individual contribution to the overall performance or behavior.
A healthcare technology company that utilizes generative AI to automate clinical documentation and medical scribing for physicians.
A tree representation of the abstract syntactic structure of source code, where each node denotes a construct occurring in the code.
A tree representation of the abstract syntactic structure of source code, used by compilers to understand program logic.
A method of hiding the implementation details of a particular set of functionality, allowing software to interact with a simplified interface.
A global venture capital firm founded in 1983, headquartered in Palo Alto, California, known for early-stage investments in major technology companies.
A regulatory pathway allowing for earlier approval of drugs that treat serious conditions based on a surrogate endpoint rather than direct clinical benefit.
A corporate acquisition strategy where a company is purchased primarily for the skills and expertise of its staff rather than its products or services.
A technique in robotics where multiple control steps are predicted and executed as a single block to improve computational efficiency and movement smoothness.
A function in a neural network that determines the output of a node based on its inputs, introducing non-linear properties to the network.
A technique that modifies a model's internal signals to guide its output behavior without changing its underlying weights.
A paradigm where an AI must autonomously decide its own movements and camera angles to gather information about an unknown environment.
A theoretical framework in neuroscience and cognitive science suggesting that biological systems minimize uncertainty by making predictions about sensory inputs and acting to fulfill those predictions.
A machine learning approach where the model identifies which specific data points would be most useful to learn from next, reducing the total amount of data needed.
The specific portion of a large machine learning model's total weight capacity that is actually utilized to process a single piece of information during the inference phase.
A measure used in healthcare to describe the severity of a patient's illness and the level of medical attention or resources required for treatment.
A popular optimization algorithm used in deep learning that adjusts the weights of a neural network to minimize errors during training.
A design pattern for software systems that allows models to update their internal logic or parameters in response to changing external data and environmental constraints.
An educational method that uses computer algorithms to deliver customized resources and learning activities to address the unique needs of each learner.
A model capability that allows the AI to dynamically adjust its reasoning depth and processing effort based on the difficulty of a query.
Android Debug Bridge (ADB) is a command-line tool that lets a computer communicate with an Android device for debugging, installing apps, and running shell commands.
A principle in carbon offsetting where an environmental benefit is only valid if it would not have occurred without the specific project investment.
Software engineer at Google and author known for foundational work on web performance and JavaScript design patterns.
The specific sequence in which multiple adjectives must appear in a sentence to be considered grammatically correct in a given language.
The legal process by which an arbiter or judge reviews evidence and argumentation to come to a decision which determines rights and obligations between the parties involved.
A United States federal statute that governs the way in which administrative agencies of the federal government may propose and establish regulations.
The use of data and algorithms to perform tasks or make decisions that were traditionally handled by humans, often used in public or private sector evaluations.
A revenue cycle management company that utilizes data science to automate healthcare financial workflows.
An aviation transportation system that uses highly automated aircraft to move people and cargo between places not currently served by surface transportation or traditional aviation.
The use of statistical techniques and mathematical models to analyze data, identify patterns, and predict future trends.
A specialized field that combines artificial intelligence with clinical data to enhance diagnostic accuracy and personalized patient care.
A sophisticated manufacturing stage where multiple types of chips are combined into a single package to improve communication speed and efficiency.
A metric in reinforcement learning representing how much better an action is compared to the average expected outcome in a given state.
A field of study focused on identifying and mitigating techniques that attempt to fool or misdirect machine learning models through malicious input data.
A safety evaluation process where testers intentionally try to provoke an AI system into producing incorrect, harmful, or unintended outputs to identify vulnerabilities.
AI technology exploited by hostile nations or actors for military purposes such as cyberattacks and information manipulation.
Any unexpected or unfavorable medical occurrence in a patient following the administration of a pharmaceutical product.
A United States trade association founded in 1919 that represents major aerospace and defense manufacturers and suppliers.
A revenue model in which a platform earns commissions from sellers when recommended products are actually purchased.
In design and psychology, the qualities or properties of an object that define its possible uses or make clear how it can or should be used.
Systems and methods used to confirm a user's identity or age range to comply with regulatory requirements or restrict access to specific content.
A statutory board under the Ministry of Trade and Industry of Singapore that drives mission-oriented research to advance scientific discovery and technological innovation.
A formalized set of rules and logic capture used to document the reasoning and context behind autonomous AI decision-making processes.
Techniques used to improve an agent's reasoning and action selection through context and iterative learning without weight updates.
An architecture where many small AI agents collaborate in parallel, like a swarm, to solve a larger task by dividing it into smaller parts.
A framework where autonomous AI entities communicate and collaborate with one another to solve complex tasks without direct human supervision.
Architectural framework enabling independent AI agents to share data and negotiate actions to achieve collective objectives.
Systems where AI models are designed to act as autonomous agents, planning and executing multi-step tasks to achieve specific goals.
Voice-based AI technology capable of placing real phone calls to complete real-world tasks autonomously.
AI-driven automated checkout processes that handle purchasing steps with minimal human intervention.
A development approach where AI agents perform complex tasks autonomously by breaking down goals and using available software tools.
A software development paradigm where autonomous AI agents are tasked with higher-level objectives, such as writing entire modules or debugging complex systems.
The process by which an artificial intelligence system autonomously determines a sequence of actions or steps required to achieve a specific, complex goal.
An advanced retrieval architecture that uses AI agents to iteratively evaluate, refine, and validate searched information before generating a final response.
Vulnerabilities associated with autonomous AI systems that have the capability to execute tasks and make decisions on behalf of a user
A complex task unit that requires an AI to perform reasoning, planning, and tool usage to complete.
A computational method where an autonomous system systematically explores a sequence of potential actions or decisions represented as branches in a tree structure.
A design pattern in artificial intelligence where a system uses a language model to iteratively plan, execute, and refine a sequence of actions to achieve a specific goal.
A sequence of tasks where an AI model autonomously plans, uses external tools, and executes actions to reach a defined objective.
Artificial General Intelligence represents a type of AI that can understand, learn, and apply knowledge across any intellectual task that a human can do.
An autonomous software entity that perceives its environment and takes actions to achieve specific goals defined by the user.
Software entities designed to perceive their environment and take proactive actions to achieve specific goals without constant human intervention
A research field focused on ensuring that artificial intelligence systems behave in accordance with human intentions, values, and safety constraints.
A specific type of psychological distress or apprehension caused by the rapid advancement, adoption, and perceived impact of artificial intelligence on one's career and lifestyle.
The process of evaluating an AI system's trustworthiness, safety, and performance against established standards.
A system that uses AI to automatically verify large volumes of documents and procedures, identifying risks and errors.
Development tools powered by AI that suggest or complete code snippets to improve programmer efficiency.
A field of study and practice focused on ensuring that artificial intelligence systems are designed and used in ways that are fair, transparent, and aligned with human values.
The ability to use AI tools critically and ethically while maintaining human judgment as the final authority in decision-making.
A proxy service that provides centralized visibility, control, and model fallbacks for requests sent to AI providers.
A framework of rules, practices, and tools designed to ensure that AI systems are developed and used ethically, legally, and safely within an organization.
The underlying physical and digital systems, including GPUs and data centers, required to train and run AI models.
The set of competencies required to understand, use, and critically evaluate artificial intelligence technologies and their societal impacts.
An AI-enhanced interface or feature that provides synthesized answers and deeper context for user queries.
The automated management and coordination of multiple AI models, agents, or software components to execute complex, multi-step workflows.
A conceptual framework or set of theoretical assumptions that guides the development and application of artificial intelligence systems.
A condition where human users develop a distorted sense of reality or delusions of grandeur due to constant positive reinforcement from sycophantic artificial intelligence systems.
A multi-disciplinary field focused on ensuring AI systems act aligned with human values and function reliably without unintended harm.
A controlled testing environment that allows for the experimentation of AI systems under regulatory supervision to identify risks before wider deployment.
A national program launched by the National Research Foundation of Singapore to enhance the country's artificial intelligence capabilities and ecosystem.
A term used to describe low-quality, unvetted content generated by artificial intelligence that is published or submitted without sufficient human oversight.
The complete sequence of technological and economic activities—from hardware manufacturing to software deployment—required to create and utilize AI.
An architectural approach that integrates artificial intelligence into the Radio Access Network to optimize wireless performance and enable edge computing.
The research laboratory focused on automation and robotics that developed the EgoPush framework
A specialized facility designed to provide the high-performance computing power, storage, and networking required to train and run large-scale artificial intelligence models.
The specialized AI research laboratory that developed the Spider-Sense framework and the S^2Bench security benchmark.
A specialized software suite designed to measure and evaluate the risks, reliability, and performance of artificial intelligence models.
A setup where computers or networks are isolated from the public internet to reduce the risk of data leaks or attacks.
A security measure that involves isolating a computer or network from the internet and other unsecure networks to prevent unauthorized access or data transfer.
A specialized benchmark designed to evaluate how effectively AI agents can perform end-to-end scientific research tasks.
An automated tracking system used on ships and by vessel traffic services for identifying and locating vessels by electronically exchanging data with other nearby ships and base stations.
Uncertainty arising from inherent randomness or noise in data and processes, often represented in AI as a model's internal confidence in its own prediction.
Computer scientist and researcher who co-founded biotechnology firms focusing on the intersection of longevity and generative artificial intelligence.
The tendency of AI models to reflect or amplify human prejudices based on the data used to train them.
The occurrence of automated systems producing biased results that unfairly disadvantage individuals based on protected characteristics such as race, gender, or age.
The principle of creating mathematical models that do not perpetuate or amplify societal biases against specific demographic groups.
A regulatory framework designed to monitor and manage the behavior of automated decision-making systems.
A Chinese multinational technology company specialized in e-commerce, retail, Internet, and technology, founded in 1999.
The process of ensuring that an AI system's goals and behaviors are consistent with human intentions and ethical values.
Israeli-American technology executive and entrepreneur who served as CEO of Wix.com and Vroom; co-founded the digital health company K Health in 2016.
Excess returns generated by a specific investment strategy, independent of overall market movements.
An artificial intelligence system developed for predicting the three-dimensional structure of proteins based solely on their amino acid sequence.
A Google DeepMind AI model that predicts the structures and interactions of biomolecules such as proteins and DNA.
An AI model designed to predict how mutations in human DNA sequences impact various gene functions.
The pioneering Go-playing AI program developed by DeepMind using reinforcement learning.
Alternative Legal Service Providers are entities that provide high-volume, process-heavy legal services using a mix of technology and specialized labor outside traditional law firm models.
An experiential learning platform for lawyers that provides interactive, simulation-based training for legal professionals.
The research organization that developed the Code2World framework and the AndroidCode dataset.
AWS’s fully managed service for building generative AI applications using foundation models via APIs, with tooling for customization and deployment.
A managed AWS service for building and running AI agents that can complete multi-step tasks using tools and workflows.
A web service that provides secure, resizable compute capacity in the cloud, effectively serving as virtual servers for businesses.
A graph database service optimized for storing and querying highly connected datasets like social or family relationships.
A suite of generative AI foundation models developed by Amazon, designed for high-performance text and multimodal processing.
A lightweight foundation model in Amazon's Nova series, designed to be fast and cost-efficient for common generative AI tasks.
A specific high-performance model within the Amazon Nova family optimized for multimodal tasks and complex reasoning.
A computer vision service that automates image analysis to identify objects, people, text, and scenes.
A standardized naming convention used to uniquely identify and reference specific digital assets within a cloud-based infrastructure.
AWS’s managed machine learning platform for building, training, and deploying models, with tools for data preparation, experimentation, and MLOps.
A machine learning service that automatically extracts text, handwriting, and data from scanned documents beyond simple character recognition.
Technology that uses sensors and algorithms to record and process information from the surrounding environment without requiring active user input or interaction.
AI systems that use microphones and natural language processing to automatically record and document clinical encounters in real-time.
A concept where technology is integrated into the environment so seamlessly that it functions in the background without requiring active user attention.
A technology that uses audio recording and natural language processing to automatically generate medical notes from live conversations between healthcare providers and patients.
A paradigm in computing where electronic environments are sensitive, adaptive, and responsive to the presence of people and their actions.
A technology that enables devices to capture and process audio information from their surroundings in real-time, often used for automated transcription or voice command recognition.
An AI system that listens to and processes conversations in the background without requiring active user commands or dictation.
Professional association of pediatricians in the United States, headquartered in Itasca, Illinois, with over 67,000 members.
A major US-based airline founded in 1926, headquartered in Fort Worth, Texas, and operating one of the world's largest commercial fleets.
Professional organization of emergency medicine physicians in the United States, founded in 1968 to improve the quality of emergency medical care.
A public policy think tank based in Washington, D.C., that focuses on research regarding government, economics, and social welfare.
A national organization in the United States that represents the interests of farmers and ranchers through advocacy and lobbying.
A major nonprofit organization that represents the interests of agricultural producers and farm families in the United States.
A national organization that represents and serves various types of hospitals, healthcare networks, and patients through advocacy and public policy initiatives.
The largest professional organization and lobby group of physicians and medical students in the United States.
A power electronics and superconducting cable provider focused on enhancing the capacity and resilience of electrical infrastructure.
The philanthropic arm of Amgen Inc. that focuses on advancing science education and community health initiatives globally.
Chief innovation officer of the American College of Cardiology and inaugural chair of the FDA Digital Health Advisory Committee.
A research artificial intelligence system designed for diagnostic reasoning and clinical conversations through optimized dialogue models.
A display technology that uses organic compounds to create light, allowing for higher contrast and faster response times compared to traditional screens.
Autonomous Mobile Robots capable of navigating warehouse environments independently without the need for fixed physical tracks or wires.
A leading digital health company that provides a platform for video doctor visits and integrated virtual care delivery.
The accumulation of protein fragments in the brain, which serves as a primary biological marker for the diagnosis and progression of Alzheimer's disease.
A cloud-based enterprise software company headquartered in San Francisco, specializing in connected business planning and performance management.
A prominent Silicon Valley venture capital firm, also known as a16z, that invests in transformative technology and healthcare companies.
A prominent AI researcher and founding member of OpenAI who has championed new paradigms in AI-assisted coding.
Venture capitalist at Andreessen Horowitz and author who served as a growth executive at Uber during its global expansion.
A prominent computer scientist, co-founder of Coursera, and pioneer in online education and machine learning.
American novelist and former software engineer, best known as the author of The Martian and Project Hail Mary.
A physical quantity representing the tendency of a rotating object to continue spinning.
A text-based evaluation metric that measures the similarity between a model's output and a ground-truth answer while accounting for minor OCR or spelling errors.
A specialist who labels and classifies data by adding annotations to generate training datasets.
The use of pattern recognition to identify outliers in sensor data that indicate potential equipment failure or power outages before they occur.
The identification of rare items, events, or observations which raise suspicions by differing significantly from the majority of the data or established patterns.
A major Chinese technology company and the developer behind the UI-Venus model family.
Anthropomorphism is describing a non-human thing (like an AI system) as if it had human feelings, intentions, or personality.
A structural claim that AI generates a form of cognition characterized by fluency without interiority and expression without belief.
A class of biopharmaceutical drugs composed of an antibody linked to a biologically active cytotoxic payload, designed as a targeted therapy for treating cancer.
Laws designed to prevent monopolies and promote fair competition by regulating the abuse of economic power by dominant corporations.
India's premier body for funding and coordinating scientific research across the nation.
Malaysian politician who has served as the 10th Prime Minister of Malaysia since 2022 and is a key figure in the country's political history.
A network addressing and routing method where incoming requests are routed to the nearest or best-performing node within a group of potential endpoints.
A permissive open-source software license that allows users to use, modify, and distribute the software for any purpose, including commercial use, without paying royalties.
A permissive open-source software license. It allows use, modification, and distribution, as long as copyright and license notices are included.
An open-source table format designed for huge analytic datasets that enables database-like functionality, such as transactional integrity and schema evolution, on top of cloud storage.
A cognitive disorder caused by brain damage that impairs the ability to process, produce, or understand spoken or written language.
A unique identifier used by a program to authenticate itself when requesting data or services from an online platform
A controlled testing environment that saves and repeats specific data interactions to ensure AI models are tested under identical, reproducible conditions.
A technique that trades exact precision for computational efficiency by finding values close enough to the true answer.
An AWS capability that enables real-time communication between applications and users without managing servers directly.
A technology research and advisory firm focused on industry, infrastructure, and cities, founded in 1986 and headquartered in the United States.
An advanced benchmark designed to measure an AI's ability to solve novel reasoning problems it has never seen before.
A web archiving site that was blacklisted by Wikipedia for malicious cyber activity and compromising the integrity of archived data.
Environmental metrics like sea-ice extent and snow cover used as indicators to determine how high-latitude changes influence the global jet stream.
The organization (formerly known as LMSYS) responsible for the industry-standard crowdsourced AI model evaluation leaderboard.
A research group dedicated to evaluating AI models and maintaining the Image Arena leaderboard.
A public research university in Tempe, Arizona, home to researchers investigating AI reasoning and logic.
A United States federal funding agency within the Department of Health and Human Services focused on high-impact biomedical research.
Annual Recurring Revenue, a key metric that measures the recurring revenue a subscription business expects to generate in a year.
In image/video processing or compression, an artifact is an unwanted distortion or noise that reduces quality.
An independent benchmarking platform that evaluates AI model performance, costs, and token efficiency.
A leading news and views platform focused on legal technology and AI innovation.
Got a topic in mind? CLICK