AllABCDEFGHIJKLMNOPQRSTUVWXYZAActivation FunctionA mathematical equation within a neuron that determines its output by processing the weighted sum... Learn More → AI and Knowledge Worker ProductivityStudies suggest knowledge workers using generative AI for tasks "inside the frontier" of AI capabilities... Learn More → AI as WavesArtificial intelligence is described as a massive set of waves that are upon us, suggesting... Learn More → AI ComputeIn AI context, often used as a noun referring to the amount or existence of... Learn More → AI Effect (AI Paradox)A phenomenon stating that once a problem thought to require intelligence is solved by a... Learn More → AI LiteracyAI literacy is the ability to recognise, grasp, use, and critically assess artificial intelligence technologies... Learn More → AI WinterA period of decreasing interest and funding in AI research and work. Characterised by a... Learn More → AlexNetAn eight-layer neural network created by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton. It won... Learn More → Algorithmic Bias (Bias and Fairness)Consistent, systematic errors in AI systems creating unfair outcomes by unjustly favoring one group. Stems... Learn More → Alignment ProblemThe challenge of ensuring that the goals of an AI system are aligned with human... Learn More → AlphaGoDeveloped by Google DeepMind. Defeated world Go champion Lee Sedol in 2016 using deep neural... Learn More → Ancient MythsSince ancient times, humans have been fascinated by the idea of intelligent machines, exemplified by... Learn More → Artificial Intelligence (ChatGPT)OpenAI's ChatGPT (GPT-4) defined AI as referring "to the simulation of human intelligence in machines... Learn More → Artificial Intelligence (McCarthy)Defined by John McCarthy, who coined the term in 1955, as "the science and engineering... Learn More → Artificial Intelligence (Oxford)The Oxford English Dictionary defines "artificial intelligence" as "The capacity of computers or other machines... Learn More → Artificial Intelligence (US Congress)The National Artificial Intelligence Initiative Act of 2020 defined "artificial intelligence" as "a machine-based system... Learn More → Artificial NeuronA computational model inspired by biological neurons. Receives digital inputs, multiplies by weights, sums them... Learn More → Ascent of Neural NetworksWhile present from the beginning (Dartmouth workshop), neural networks fell out of favour, but research... Learn More → Automation (Home)AI is used to automate household chores and tasks (e.g., robot vacuums, automated dishwashing) and... Learn More → Automation (Workplace)AI is used to automate tedious, repetitive tasks and business processes (e.g., document processing, legal... Learn More → BBackpropagationA method for training DNNs. Involves a forward pass (input to output prediction) and a... Learn More → Biological NeuronA nerve cell in the human brain composed of a nucleus, axon, and dendrites, communicating... Learn More → Burden of EaseRelates to the potential loss of character building and connection to oneself and the world... Learn More → CChatGPTReleased by OpenAI on November 30, 2022. An AI chatbot using large language models (LLMs)... Learn More → ClusteringAn unsupervised ML algorithm used to find groups (clusters) in unlabeled data where members are... Learn More → Combating DisinformationAI is also used to scan content, identify signs of disinformation or fake news, and... Learn More → Convolutional Neural Network (CNN)A type of DNN primarily used in computer vision to identify and classify objects in... Learn More → Credit Assignment ProblemThe challenge in training DNNs to figure out how much each weight in the hidden... Learn More → DDartmouth WorkshopHeld in Hanover, New Hampshire during the summer of 1956, organised by John McCarthy. Widely... Learn More → Data LiteracyData literacy is defined as the ability to read, understand, create, and communicate data as... Learn More → Data Quality and BiasReal-world data used for training AI is often "dirty" (missing values, errors) and reflects societal... Learn More → Deep Learning (DL)A special branch and subset of machine learning involving deep neural networks (DNNs). It's contained... Learn More → Deep Neural Network (DNN)A neural network with two or more hidden layers between the input and output layers.... Learn More → EEconomic InequalityAI has the potential to deepen inequality within and between countries. Workers who can harness... Learn More → Economic/Financial BenefitsAI can drive productivity gains for companies and economies by automating tasks and enhancing existing... Learn More → ELIZACreated in 1966 by MIT professor Joseph Weizenbaum. It was one of the first "chatterbots"... Learn More → Environmental Factors Fueling MLMachine learning's ascendance has been fueled by dramatic increases in AI compute (computational power/speed, including... Learn More → Environmental ImpactAI training and operations require massive energy, burdening the environment. However, AI can also be... Learn More → Existential ThreatA controversial, hypothetical concern that AI's intelligence could grow exponentially beyond human control (technological singularity),... Learn More → Expert SystemsAI programs popular in the 1980s using predefined rules and knowledge bases in specific fields... Learn More → FFeedforward Neural NetworkA type of neural network where connections only go forward, from input to output layers,... Learn More → First AI WinterMid- to late 1970s, often attributed to overpromising and underdelivering on early AI predictions. Learn More → GGeneral Problem Solver (GPS)A rules-based symbolic AI program developed by Newell, Simon, and Shaw in 1957. Intended to... Learn More → General Purpose AI (GPAI)A relatively new term defined in the EU's AI Act for AI models displaying significant... Learn More → Generative Adversarial Network (GAN)AI architecture introduced in 2014. Consists of a generator neural network that creates content (fakes)... Learn More → Golden Age of AIThe first two decades after the Dartmouth workshop (1956-1974), characterised by enthusiasm and optimism for... Learn More → GPUs (Graphical Processing Units)Special computer chips originally for graphics, repurposed for training machine learning models, especially deep neural... Learn More → Gradient DescentA mathematical technique used in backpropagation to find the direction and amount to change weights... Learn More → HHidden LayerAny layer of neurons in a neural network located between the input layer and the... Learn More → HyperparameterAspects of a neural network set by humans before training (e.g., number of layers/neurons, activation... Learn More → IImageNetA massive database of labeled photographs built by Fei-Fei Li and collaborators starting in 2006... Learn More → Information and PropagandaAI may spread disinformation and propaganda, including deepfakes, potentially affecting public opinion, voting behaviour, and... Learn More → JJob DisplacementNew technologies like AI can eliminate some jobs, augment others, and create new ones, leading... Learn More → LLarge Language Model (LLM)A type of deep neural network trained on massive text data to process and generate... Learn More → LeNetProposed in 1989 by Yann LeCun. A convolutional neural network architecture employing backpropagation to recognise... Learn More → Logic TheoristPresented at the 1956 Dartmouth workshop by Allen Newell, Herbert A. Simon, and Cliff Shaw.... Learn More → MMachine Learning (ML) BasicsML is an umbrella term for statistical algorithms applied to data to learn patterns and... Learn More → NNeocognitronInvented in 1979 by Kunihiko Fukushima. A multilayered neural network trained to recognise features in... Learn More → Neural NetworkAn AI program containing many artificial neurons arranged in layers and connected. A simple type... Learn More → OOverfittingA problem in ML where a model learns noise and irrelevant details in the training... Learn More → PPerceptronDeveloped in 1957 by Frank Rosenblatt. The first actual implementation of an artificial neural network,... Learn More → Potential Copyright InfringementGenerative AI models trained on large amounts of internet data (including copyrighted works) can produce... Learn More → RReinforcement Learning (RL)One of the three major forms of ML. An AI agent learns to make decisions... Learn More → Reinforcement Learning from Human Feedback (RLHF)A training method used for finetuning LLMs like GPT, where human evaluators assess and score... Learn More → SSecond AI WinterStarting around 1990 and extending to the early 2000s, coinciding with the decline of interest... Learn More → Self-Attention MechanismAn advancement used in transformer networks allowing simultaneous consideration of every token in a sequence... Learn More → Self-supervised LearningA form of ML where the system itself creates the labels for training data, rather... Learn More → Semi-supervised LearningA form of ML combining supervised and unsupervised learning, using a small amount of labeled... Learn More → Strong AI (AGI)Also called general AI or artificial general intelligence (AGI), it's a hypothetical type of AI... Learn More → Subsymbolic AIThe other main branch of AI, also called connectionist AI. It focuses on learning and... Learn More → Supervised LearningOne of the three major forms of ML. Goal is to teach a model to... Learn More → Sustainable Development Goals (SDGs)AI can contribute to achieving UN SDGs, such as managing energy grids more efficiently, optimising... Learn More → Symbolic AIOne of the two main branches of AI, also called classic AI. It involves explicit... Learn More → TTechnological SingularityA hypothetical future point where AI self-improves uncontrollably, leading to an intelligence explosion dwarfing human... Learn More → TokenA chunk of text (word, fragment, letter, phrase) processed by transformers in LLMs. Learn More → Training DNNsInvolves methods like backpropagation and gradient descent, typically as part of supervised learning, to adjust... Learn More → TransformerA specific type of neural network architecture introduced in 2017, using a self-attention mechanism that... Learn More → Turing TestProposed by Alan Turing in his 1950 paper "Computing Machinery and Intelligence". It's an "imitation... Learn More → UUnderfittingA problem in ML where a model is too simple and doesn't learn enough relevant... Learn More → Unsupervised LearningOne of the three major forms of ML. Learns from patterns in unlabeled data without... Learn More → WWeak AI (ANI)Also called narrow AI or artificial narrow intelligence (ANI), it refers to AI designed to... Learn More → Load More
