Skip to Main Content

Guide to Artificial Intelligence in Human Resources Terminology

Mindy Honcoop
on November 09, 2023

A Glossary of 305 (Updated 05/09/24) AI in HR Terms and Definitions

Artificial intelligence (AI) and natural language (NL) technologies play a crucial role in the world of business today. Yet, we understand that these topics can seem daunting and intricate to many. Fear not! Everyone should have the chance to engage in this conversation. That’s why we’ve put together a friendly glossary of AI and NL terms to make things easier for you.  This list of terms encompasses the keywords and phrases fundamental to understanding natural language and artificial intelligence technologies. With these, you can confidently embark on your journey. 

Navigating AI Terminologies Inside This Artificial Intelligence Article.

A (38) | B (13) | C (29) | D (20) | E (19) | F (12) | G (9) | H (6) | I (7) | J (1) | K (3) | L (13) | M (16) | N (11)
| O (6) | P (20) | Q (1) | R (17) | S (18) | T (16) | U (5) | V (3) | W (1) | X (1) | Y (4) | Z (1)

SPECIAL: AI (15) for HR and Recruiting Terms

A (38) – AI: Automating your workload, one algorithm at a time.

A/B TestingA controlled, real-life experiment designed to compare two variants of a system or a model, A and B.
AccuracyA scoring system in binary classification. 
Active Learning (Active Learning Strategy)A special case of Semi-Supervised Machine Learning in which a learning agent is able to interactively query an oracle (usually, a human annotator) to obtain labels at new data points.
Adversarial Machine LearningA technique in machine learning where two neural networks, the generator, and the discriminator, are pitted against each other to improve the model’s performance. 
AgentsSoftware capable of performing tasks independently
AGI (Artificial General Intelligence)AI as capable as a human in any intellectual task. 
AI AssistAI Assist is an AI-powered system that supports users by understanding queries, providing information, and performing tasks. It enhances productivity and convenience in various applications like customer service and virtual assistants.
AI BiasThe presence of systematic and unfair discrimination in AI systems, often a result of biased training data or algorithms, leading to unjust or unfair outcomes.
AI ChatbotA computer program that uses artificial intelligence to engage in human-like conversations with users, often used for customer support and information retrieval.
AI CopilotAn AI-powered virtual assistant in an enterprise using advanced language models and generative AI, it offers human-like interactions, proactive notifications, and automated task resolution across multiple domains, including customer service, employee experience, and operational efficiency.
AI DiscoveryInvolves leveraging machine learning technology to extract valuable insights and patterns from large datasets. By using advanced algorithms and data-analysis techniques, businesses are able to uncover hidden correlations, trends, and anomalies, enabling data-driven decision-making and the discovery of new knowledge. 
AI EthicsThe field of study and practice that focuses on the moral and ethical implications of artificial intelligence and the development of guidelines and principles for responsible AI use. 
AI ExplainabilityThe capability of AI systems to provide understandable and transparent explanations for their decisions and actions, especially in critical applications like healthcare and finance.
AI ModelA mathematical representation of an AI system’s knowledge and capabilities, which is used for various tasks such as prediction, classification, or generation. 
AI ObservabilityRefers to the practice of monitoring, analyzing, and gaining insights into the behavior and performance of artificial intelligence systems. It involves collecting and analyzing structured data related to AI models, training processes, and deployments to ensure transparency, reliability, and effectiveness in AI operations. 
AI SafetyThe study of methods and techniques to ensure that AI systems operate safely and reliably, minimizing risks and potential harm.
AI StackThe layers and components of technology and tools that form the foundation of AI systems, including hardware, software, and algorithms.
AI Service DeskAn automated support system powered by artificial intelligence. It assists users in resolving their queries or issues by using natural language processing, machine learning algorithms, and other AI techniques, thereby improving efficiency and reducing response times.
AI Superintelligence (ASI)AI surpassing human capabilities.
AI Ticket AssistAi refers to an artificial intelligence system used in customer service or IT service management. It automates the process of ticket creation, categorization, prioritization, and routing. It also assists in resolving tickets faster by providing relevant solutions based on past data and machine learning.
AI Virtual AssistantsAn AI-based software or application mostly called, AI Virtual Assistant, provides support, performs tasks, and answers questions through natural language interactions and conversational AI technologies, often utilizing speech and image recognition and synthesis.
AlignmentEnsuring AI system goals align with human values.
AnaphoraA reference to a noun by way of a pronoun.
AnnotationThe process of tagging language data.
Area Under the Curve (AUC)A methodology used in Machine Learning to determine which one of several used models has the highest performance.
Artificial Intelligence (AI)The use of technology to simulate human intelligence, either in computer programs or robotics. A field in computer science that aims to build systems that can perform human tasks.
Artificial Intelligence Operations (AIOps)Leverages AI and machine learning techniques to IT operations that enable automated monitoring, detection, and resolution of issues or anomalies—transformative impact. AIOps stands for artificial intelligence for IT operations. 
Artificial Neural Network (ANN)A system mimicking the human brain’s processing abilities. 
Augmented IntelligenceAn architecture composed of successive layers of simple connected units called artificial neurons interweaved with non-linear activation functions, which is vaguely reminiscent of the neurons in an animal brain.
Auto-classificationAutomatically classifying text using AI techniques.
Auto-completeA search functionality suggesting possible queries. 
AutoencoderA type of Artificial Neural Network used to produce efficient representations of data in an unsupervised and non-linear manner, typically to reduce dimensionality.
Automated Speech RecognitionA subfield of Computational Linguistics interested in methods that enable the recognition and translation of spoken language into text by computers. 
AutomationThe use of largely automatic equipment in a system of operation.
AnthropomorphismThe tendency to attribute human qualities to nonhumans. For example, calling a chatbot “he” or “she,” saying a chatbot wants something or saying a chatbot is trying to do something is anthropomorphizing AI. This happens often in conversations about artificial intelligence because it is often designed to sound or appear human. 
AlgorithmAn unambiguous specification of a process describing how to solve a class of problems that can perform calculations, process data, and automate reasoning.
Active Learning (Active Learning Strategy)A special case of Semi-Supervised Machine Learning in which a learning agent is able to interactively query an oracle (usually, a human annotator) to obtain labels at new data points.
Actionable IntelligenceInformation that can support decision-making.

B (13) – Break Room: The sacred place for coffee breaks and snack raids.

  • Backpropagation (Backpropagation Through Time): A method used to train Artificial Neural Networks to compute a gradient that is needed in the calculation of the network’s weights. 

  • Bard: Developed by Google, an AI-powered chatbot that can respond to user questions on any subject. It’s designed to simulate human conversations using natural language processing and machine learning 

  • BERT (Bidirectional Encoder Representation from Transformers): Google’s technology for large-scale pretrained models. 

  • Behavior Trees: A tool used in AI programming to model the behavior of characters or entities in video games or simulations. They use a tree-like structure to represent decision-making processes. 

  • Big Data: A term used to describe extremely large and complex datasets that may be analyzed to reveal patterns, trends, and associations. Big data plays a crucial role in AI and machine learning. 

  • Binary Classification: A machine learning task where the goal is to categorize data into one of two classes, such as “yes” or “no,” “spam” or “not spam,” etc. 

  • Bioinformatics: The application of AI and computational techniques to analyze and interpret biological data, including DNA sequences, protein structures, and more. 

  • Bots: Autonomous software programs that perform tasks automatically, often in the context of chatbots, customer service, or automation. 

  • Blockchain: A decentralized and distributed ledger technology that can be used to secure and verify transactions and data, with potential applications in AI, especially for data integrity and security. 

  • Bias: Assumptions AI models make about data. 

  • Bias-Variance Tradeoff: A conflict arising when data scientists try to simultaneously minimize bias and variance, which prevents supervised algorithms from generalizing beyond their training set. 

  • Bounding Box: The smallest (rectangular) box fully containing a set of points or an object. 

  • Bayes’s Theorem: A famous theorem used by statisticians to describe the probability of an event based on prior knowledge of conditions that might be related to an occurrence. 

C (29) Commute: That daily race to work where red lights are your worst enemy.

Causal InferenceThe process of determining whether one event or variable causes another, often used in AI for understanding cause-and-effect relationships in data.
CategorizationAssigning categories to documents.
CategoryA label describing document content.
Category TreesViewing rule-based categories. 
Chain of ThoughtSequence of reasoning steps in AI decision-making.
ChatbotA computer program or an AI designed to interact with human users through conversation.
ChatGPTDeveloped by OpenAI, a large AI language model generating human-like text.
CLIP (Contrastive Language–Image Pre-Training)Connecting text and images.
ClassificationThe task of approximating a mapping function from input variables to discrete output variables, or, by extension, a class of Machine Learning algorithms that determine the classes to which specific instances belong.
Collaborative RoboticsThe collaboration between robots and humans in shared workspaces, often used in manufacturing and logistics, with potential applications in AI for robot control and coordination.
Collaborative FilteringA method used in the context of recommender systems to make predictions about the interests of a user by collecting preferences from a larger group of users. 
Computer Speech RecognitionThe ability of a computer system to convert spoken language into written text, allows for machine translation of voice commands and dictation. 
Computer VisionThe field of Machine Learning that studies how to gain high-level understanding from images or videos.
Confidence IntervalA type of interval estimate that is likely to contain the true value of an unknown population parameter. The interval is associated with a confidence level that quantifies the level of confidence of this parameter being in the interval.
Concept DriftThe phenomenon where the statistical properties of data change over time, which can affect the performance of AI models.
Constitutional AITrains AI systems to align with a set of values or principles as defined in a constitution. This approach was developed by AI startup Anthropic. 
ContentContainers of information for training data or generative AI. 
Content Enrichment or EnrichmentExtracting meaningful information from text-based documents using advanced techniques. 
Contextual UnderstandingThe ability of AI systems to comprehend and interpret information based on its context, often used in natural language processing and understanding. 
Controlled VocabularyCurated collection of words and phrases. 
Conversational AIBuilding conversational user interfaces and chatbots. 
Conversational AI AssistRefers to an artificial intelligence system designed to simulate human-like interactions. Through natural language processing and machine learning methods, it can understand and respond to user queries in a conversational manner, improving customer service experiences on digital platforms. 
Conversational AutomationThe concept that combines the power of conversational AI and automation. It involves automating and streamlining interactions with users through chatbots or virtual assistants, enabling efficient and effective communication. By leveraging natural language understanding and automated processes, conversational automation enhances customer service, information retrieval, and task completion in a conversational manner.
Conversational SearchAn approach to information retrieval that utilizes conversational AI techniques. Instead of traditional keyword-based queries, users can interact with search systems using natural language, similar to having a conversation. By understanding user intent and context, conversational search systems provide more relevant and personalized search results, improving the overall search experience. 
Convolutional Neural Network (CNN)A class of Deep, Feed-Forward Artificial Neural Networks, often used in Computer Vision. 
CorpusThe entire set of language data for analysis.
Curse of Dimensionality:Phenomena that arise when analyzing and organizing data in high-dimensional spaces due to the fact that the more the number of dimensions increases, the sparser the amount of available data becomes.
Custom/Domain Language ModelA model built for a specific organization or industry.
Concurrent TrainingA training method that allows AI models to learn from multiple sources of data or tasks simultaneously, improving their overall performance.

D (20) – Dilbert Principle: When incompetence rises to the top, just like in the comics.

  • Data (Structured Data, Unstructured Data, Data augmentation): The most essential ingredient to all Machine Learning and Artificial Intelligence projects. 

  • Data Augmentation: Increasing data diversity by modifying existing data. 

  • Data Discovery: Uncovering data insights and delivering them to users. 

  • Data Drift: Changes in the distribution of input data over time. 

  • Data Extraction: Collecting data from various sources. 

  • Data Ingestion: Obtaining and restructuring data from multiple sources.
  • Data Labelling: Marking data to make it recognizable by machines. 

  • Data Scarcity: Lack of data affecting predictive analytics. 

  • Data Validation: The process of checking the quality of data before using it to develop and train AI models. 

  • Data Mining: The process of searching for data and looking for patterns within a large data set to extract useful information. Names of entities in the data set are one example of data that mining processes can extract. 

  • Dall-E: OpenAI’s AI-powered image generator. Users submit a text prompt, and the AI tool generates a corresponding image. 

  • Decision Tree: A category of Supervised Machine Learning algorithms where the data is iteratively split in respect to a given parameter or criteria. 

  • Deep Blue: A chess-playing computer developed by IBM, better known for being the first computer chess-playing system to win both a chess game and a chess match against a reigning world champion under regular time controls. 

  • Deep Learning: Machine learning based on neural networks. 

  • Deep Learning Models: A subfield of machine learning that utilizes artificial neural networks with multiple layers to process and analyze complex patterns and structured data, mimicking the human brain’s structure and function. 

  • Deepfake: A convincing AI-generated image, audio, or video hoax. They can be entirely original content that shows someone doing or saying something they didn’t do or say. They can also depict fake news events. A deepfake of the Pentagon exploding went viral in May 2023 and had a tangible effect on the stock market. 

  • Deep Learning (Deep Reinforcement Learning): A broader family of Machine Learning methods based on learning data representations, as opposed to task-specific algorithms. Deep Learning can be supervised, semi-supervised, or unsupervised. 

  • Diffusion: Technique for generating data with added noise. 

  • Dimensionality Reduction: The process of reducing the number of random variables under consideration by obtaining a set of principal variables. Also see Feature Selection. 

  • Double Descent: Phenomenon in machine learning model performance

E (19) – Email Overload: The never-ending stream of digital paperwork.

Edge ComputingA computing paradigm that brings AI and data processing closer to the source of data generation, reducing latency and enabling real-time analysis and decision-making on devices or at the network edge.
Embedding (Word Embedding)Data representation in a new form.
Emotion AI (aka Affective Computing)Analyzing the emotional state of a user.
Encoder and Decoder NetworksThese are types of deep neural network architectures whose job it is to convert a given input, say text, into a numerical representation, such as a fixed length set of numbers (encoder), and also convert these numbers back to a desired output (decoder). They are very commonly used in natural language processing tasks such as machine translation. 
EntityAny noun, word, or phrase in a document that refers to a concept.
Ensemble MethodsIn Statistics and Machine Learning, ensemble methods use multiple learning algorithms to obtain better predictive performance that could be obtained from any of the constituent learning algorithms alone. Unlike a statistical ensemble in statistical mechanics, which is usually infinite, a machine learning ensemble consists of only a concrete finite set of alternative models but typically allows for a much more flexible structure to exist among those alternatives.
EntropyThe average amount of information conveyed by a stochastic source of data.
EpochIn the context of training Deep Learning models, one pass of the full training data set. 
Environmental, Social, and Governance (ESG)Pertains to societal impact and accountability. 
Embodied AgentsEmbodied agents, also referred to as embodied AI, are AI agents with a physical body that perform specific tasks in the physical environment. 
EmergenceEmergence describes capabilities that arise in AI systems unpredictably as they become more complex. A system’s emergent properties are not observable in its individual parts.
EntityAny noun, word, or phrase in a document that refers to a concept. 
Environmental, Social, and Governance (ESG)Pertains to societal impact and accountability. 
ETL (Entity Recognition, Extraction)Identifying relevant entities in a document.
EU AI ActRegulatory framework for responsible AI deployment in a way that doesn’t conflict with data privacy rights.
Expert SystemExpert systems are AI systems used to simulate the judgment or behavior of a human expert.
Explainable AI /Explainability:AI approach making algorithms transparent and understandable. 
Expert SystemsAI solving domain-specific problems
Extraction or Keyphrase ExtractionDescribing main ideas in text

F (12) – Flair: The more, the better—unless it’s a stapler in Jell-O.

  • Feature (Feature Selection, Feature Learning): A variable that is used as an input to a model. 

  • Feature Learning: An ensemble of techniques meant to automatically discover the representations needed for feature detection or classification from raw data. 

  • False Positive: An error due to the fact a result did reject the null hypothesis when it shouldn’t have.
  • False Negative: An error due to the fact a result did not reject the null hypothesis when it should have. 

  • Feed-Forward (Neural) Networks: An Artificial Neural Network wherein connections between the neurons do not go backward or form a cycle. 

  • F-Score: A measure of a model’s accuracy considering both the precision and the recall to compute the score. More specifically, the F-Score is the harmonic average of the precision and recall, where it reaches its maximal value at 1 (perfect precision and recall) and minimum at 0. 

  • Fréchet Inception Distance (FID): FID is a metric for evaluating the quality of images created by generative AI. 

  • Few-Shot Learning: Generalizing with few training examples. 

  • Fine-Tuned Model: Focused on a specific context or category. 

  • Fine-Tuning: Improving a pre-trained model with task-specific data.
  • Foundational Model: Baseline model pretrained on large amounts of data. 

  • Feed-Forward (Neural) Networks: An Artificial Neural Network wherein connections between the neurons do not go backward or form a cycle. 

G (9) – “Googling Your Symptoms”: The forbidden task that turns every sneeze into a terminal illness.

Garbage In, Garbage OutA principle stating that whenever the input data is flawed, it will lead to misleading results and produces nonsensical output, a.k.a. “garbage”.
General Data Protection Regulation (GDPR)A regulation in EU law on data protection and privacy for all individuals within the European Union aiming to give control to citizens and residents over their personal data. 
Genetic AlgorithmA search heuristic inspired by the Theory of Evolution that reflects the process of natural selection where the fittest individuals are selected to produce offspring of the following generation.
Generative AIA subset of artificial intelligence that focuses on creating new and original content, such as images, music, or text, using algorithms and machine learning models. It enables machines to generate creative outputs that mimic human-like patterns and styles.
Generative Adversarial Networks (GANs)A class of Artificial Intelligence algorithms used in Unsupervised Machine Learning, implemented as the combination of two Neural Networks competing with each other in a zero-sum game framework.
Generative Pre-Trained Transformer (GPT)GPTs are the AI algorithms that power some of the most well-known natural language processing and generative algorithms. GPT-3, GPT-3.5 and GPT-4 are examples of the GPT family of algorithms. They were developed by OpenAI. 
Generalized ModelA model not focused on specific use cases.
Ground TruthA piece of information obtained through direct observation as opposed to inference. 
GroundingMapping factual information in generative output.unding: Mapping factual information in generative output.

H (6) – Human Resources: The department that makes even AI’s HR software feel inadequate.

  • Hallucination: Hallucination in AI typically results from overfitting, bias, or a lack of context awareness in the machine learning model. The technologies like an image generator or large language model, that generate outputs which are not directly derived from the input data can hallucinate if not trained not to do so. 

  • Hidden Layer: A layer of neurons whose outputs are connected to the inputs of other neurons, therefore not directly visible as a network output. 

  • Human-in-the-Loop (HITL): Human-in-the-loop is a branch of artificial intelligence that leverages both human and machine intelligence to create machine learning models. In a traditional human-in-the-loop approach, people are involved in a virtuous circle where they train, tune, and test a particular algorithm. 

  • Hybrid AI: Combines multiple AI methodologies. 

  • Hyperparameter (Hyperparameter Tuning): A configuration, external to the model and whose value cannot be estimated from data, that data scientists continuously tweak during the process of training a model. 

  • Hyperparameters: Adjustable model parameters for optimal performance. 

SPECIAL: A Glossary of (15) AI for HR and Recruiting Terms

Bias Detection & MitigationAI algorithms can help identify and mitigate biases in HR processes, such as recruitment, performance evaluation, and promotion decisions. By reviewing historical data, AI can highlight potential biases and help HR professionals ensure fair and equitable practices are developed and implemented.
Candidate MatchingUses machine learning algorithms to analyze candidates’ qualifications, skills, and preferences and match them with job openings that best fit their profiles 
Employee EngagementAI employee engagement initiatives that assess employee feedback, sentiment analysis, and social media data. By understanding employee sentiments and preferences, HR professionals can tailor engagement strategies, improve employee satisfaction, and foster a positive work environment.
Employee RetentionAids in predicting and addressing employee attrition by monitoring various factors such as job satisfaction, compensation, performance, and employee sentiment. This information enables HR to implement targeted retention strategies, such as personalized interventions or career development opportunities, to increase employee loyalty and reduce turnover.
Employee WellbeingAI can assist in monitoring and promoting employee well-being by combining sentiment analysis, health-related survey responses and even data from wearable devices. By identifying patterns and trends, HR professionals can design interventions and policies to support employee well-being and create a balanced programme of automated and manual engagements to support individual mental, physical and financial wellbeing.
HR ChatbotsApplies to employee onboarding, answering frequently asked questions, and providing assistance to employees regarding policies, benefits, and HR-related inquiries. 
HR AI CopilotA virtual assistant powered by artificial intelligence that supports HR professionals in various tasks such as onboarding, employee engagement, and data analysis. It provides valuable insights, automates routine HR processes, and assists in decision-making, enhancing HR efficiency and effectiveness.
Interview IntelligenceLeverages data and technology to enhance the interview process, from preparation to evaluation. It aims to improve interview outcomes by providing insights, tools, and guidance to both candidates and interviewers, ultimately leading to more informed hiring decisions and better candidate experiences.
Learning & Development (L&D)AI-based L&D platforms offer personalized and adaptive learning experiences to employees. By looking at individual learning patterns and preferences, AI can recommend relevant training content, deliver micro-learning modules, and track employee progress, enhancing the effectiveness of training programs. 
Performance ManagementAI-powered performance management systems use data analytics to evaluate employee performance objectively. These systems generate real-time feedback, identify skill gaps, and offer development recommendations for individuals, enabling HR to make data-driven decisions regarding promotions, training, and performance improvement strategy.
Predictive AnalyticsUses AI algorithms and statistical models to analyze historical and real-time data, identifying patterns and trends to predict potential future outcomes. In HR, predictive analytics can be used to forecast possible employee turnover, anticipate hiring needs, and enhance workforce planning. 
Recruiting AutomationThe use of AI-powered tools to streamline and optimize the recruitment process. AI can assist in resume screening, candidate shortlisting, and automated scheduling of interviews, reducing manual inputs and improving overall process efficiency of HR teams. 
Screening & SelectionAutomates the initial screening and assessment of job applicants based on predetermined criteria.
Talent IntelligenceStrategic use of data and analytics to inform decisions related to an organization’s human resources, helping in talent identification, recruitment, retention, and workforce alignment with business goals.
Workforce AnalyticsThe use of AI to collate HR data, such as employee demographics, performance metrics, and engagement surveys, to gain insights into workforce trends and patterns. These insights enable HR to make informed decisions regarding talent acquisition, workforce planning, and organizational development.

I (7) – “I’ll Get to It Tomorrow”: The eternal promise of procrastination.

  • ImageNet: A large visual dataset made of 14 million URLs of hand-annotated images organized in twenty-thousand (20,000) different categories, designed for use in visual object recognition research. 
  • Image Recognition: The problem in Computer Vision of determining whether an image contains some specific object, feature, or activity. 

  • Inference: The process of making predictions by applying a trained model to new, unlabeled instances. 

  • Inference Engine: Applies logical rules to deduce information. 

  • Information Retrieval: The area of Computer Science studying the process of searching for information in a document, searching for documents themselves, and also searching for metadata that describes data, and for databases of texts, images or sounds. 

  • Insight Engines: Describe, discover, organize, and analyze data. 

  • Intelligent Document Processing (IDP) or Intelligent Document Extraction and Processing (IDEP): Reading and extracting data from unstructured and semi-structured data. 

J (1) – Jargon: Buzzwords that make you sound smart but nobody understands.

Jacobian MatrixIn the context of machine learning and neural networks, the Jacobian matrix represents the matrix of all first-order partial derivatives of a vector-valued function. It’s used for various purposes, including gradient-based optimization and sensitivity analysis.

K (3) – Keyboard Shortcuts: The secret language of the truly efficient.

  • Knowledge Engineering: Helping computers replicate human-like knowledge.field of AI that aims to emulate a human expert’s knowledge in a certain field. 

  • Knowledge Graphs: Machine-readable structures representing knowledge. 

  • Knowledge Model: A computer-interpretable model of knowledge. 

L (13) – Lunch Hour: A sacred break when office warriors refuel for the afternoon.

Labeled DataData marked for machine recognition.
LangOps (Language Operations)Workflows for language models and solutions.
Language DataWritten and spoken words in language. 
Large Language Models (LLM)Supervised learning algorithms. 
Large Language Model (LLM) AgentsOn their own, LLMs take text as input and provide more text as output. Agents are systems built on top of an LLM that give them agency to make decisions, operate autonomously, and plan and perform tasks without human intervention. Agents work by using the power of LLMs to translate high level language instructions into the specific actions or code required to perform them. There is currently an explosion of interest and development in Agents. Tools such as AutoGPT are enabling exciting applications such as “task list doers” that will take a task list as input and actually try and do the tasks for you.
Large Language Model Meta AI (LLaMA)LLaMA is an open source LLM released by Meta. 
Layer (Hidden Layer)A series of neurons in an Artificial Neural Network that process a set of input features, or, by extension, the output of those neurons.
Learning RateA scalar value used by the gradient descent algorithm at each iteration of the training phase of an Artificial Neural Network to multiply with the gradient.
Learning-to-LearnA new direction within the field of Machine Learning investigating how algorithms can change the way they generalize by analyzing their own learning process and improving on it. 
Learning-to-RankThe application of Machine Learning to the construction of ranking models for Information Retrieval systems. 
LemmaThe base form of a word.
LexiconKnowledge of word meanings in context.
Linked DataConnected knowledge stores. 

M (16) – Meetings: Where minutes are kept and hours are lost.

  • Machine Learning: The subfield of Artificial Intelligence that often uses statistical techniques to give computers the ability to “learn,” i.e., progressively improve performance on a specific task, with data, without being explicitly programmed. See also Supervised Learning, Unsupervised Learning, Reinforcement Learning 

  • Machine Learning Lifecycle Management: DevOps for Machine Learning systems. 

  • Machine Translation: A subfield of computational linguistics that studies the use of software to translate text or speech from one language to another. 

  • Model: A model is an abstracted representation of what a Machine Learning system has learned from the training data during the training process. 

  • Model Collapse: When low-quality, AI-generated content contaminates the training set for future models. 

  • Model Drift: The decay of a model’s predictive power due to changes in real-world environments. 

  • Model Parameter: Parameters in the model determined by training, necessary for making predictions. 

  • Monte Carlo: An approximate methodology that uses repeated random sampling in order to generate synthetic simulated data. 

  • Morphological Analysis: Breaking complex problems down into basic elements to gain a better understanding. 

  • Multimodal AI: Multimodal AI systems can handle input and produce output in several mediums.

  • Multimodal Models and Modalities: Language models trained on and understanding multiple data types.
  • Multi-Modal Learning: A subfield of Machine Learning aiming to interpret multimodal signals together and build models that can process and relate information from multiple types of data. 

  • Multi-Task Learning: A subfield of Machine Learning that exploits similarities and differences across tasks in order to solve multiple tasks at the same time. 

  • Multitask Prompt Tuning (MPT): An approach to configure prompts for repetitive changes. 

  • Metadata: Data that describes or provides information about other data. 

  • Metacontext and Metaprompt: Foundational instructions on how to train a model’s behavior. 

N (11) – Napping Pod: Your secret hideaway for office power naps.

Named Entity RecognitionA subtask of Information Extraction that seeks to identify and classify named entities in text into predetermined categories such as the names, locations, parts-of-speech, etc.
Natural Language Processing (NLP)The area of Artificial Intelligence that studies the interactions between compuNatural language generation (NLG)ters and human languages, in particular how to process and analyze large amounts of natural language data.
Natural Language Generation (NLG)NLG is the use of AI to produce written or spoken language from a data set. 
Natural Language Understanding (NLU)The AI’s capability to comprehend and interpret human language in a meaningful way, allows machines to extract intent, context, and relevant information from textual or spoken inputs.
Naive BayesA family of simple probabilistic classifiers based on applying Bayes’ theorem with strong independence assumptions between the features.
NeuronA unit in an Artificial Neural Network processing multiple input values to generate a single output value
Neuromorphic ComputingNeuromorphic computing is a method in which computer design is modeled after elements of the human brain. It can apply to both hardware and software. 
NeRF (Neural Radiance Fields)Creating 3D scenes from 2D images with a neural network. 
Neural NetworkAI model inspired by the brain’s structure.
Neural SearchAn advanced search technique that harnesses the power of machine learning technology, particularly deep learning algorithms, within neural networks. By utilizing these sophisticated algorithms, neural search systems can process complex search queries, enabling more precise and contextually relevant search results. This approach optimizes search engines to enhance the user experience by using the capability of machine learning and deep learning
Next-Gen IT Service Management (Next-Gen ITSM)Refers to innovative strategies in managing IT services, typically involving automation and artificial intelligence. This could include AI-powered service desks, predictive analytics, cloud services, and other advanced technologies aimed at enhancing service delivery and operational efficiency. 

O (6) – Overtime: The hours you weren’t planning to spend but end up doing anyway.

  • Omnichannel AI Support: Refers to an artificial intelligence system that provides customer service across multiple communication channels, such as email, social media, phone calls, and live chat. It ensures seamless, consistent, and personalized customer experiences, regardless of the platform used for interaction. 

  • OpenAI: An American artificial intelligence company. It conducts AI research and has developed several AI models and services in the last decade, including GPT-3, ChatGPT, and

  • Optical Character Recognition: The conversion of images of printed, handwritten, or typed text into a machine-friendly textual format. 

  • Optimization: The selection of the best element (with regard to some criterion) from some set of available alternatives. 

  • Overfitting: The fact that a model unknowingly identified patterns in the noise and assumed those represented the underlying structure; the production of a model that corresponds too closely to a particular set of data, and therefore fails to generalize well to unseen observations. 

  • Objective Function: Function to maximize or minimize during training. 

P (20) – “Please Advise”: The email signature that really means, “Help!”

ParametersInternal variables in a machine learning model.
ParsingIdentifying and assigning logical and grammatical values to text elements. 
Part-of-Speech TaggingIdentifying grammatical information about sentence elements. 
Pathways Language Model (PaLM)PaLM is Google’s transformer-based LLM, based on similar technology to GPT-3 and GPT-4. The Google Bard chatbot runs on PaLM.
Pattern RecognitionAn area of Machine Learning focusing on the (supervised or unsupervised) recognition of patterns in the data.
Personally Identifiable Information:Any piece of information that can be used on its own or in combination with some other information in order to identify a particular individual. 
PrecisionA percentage indicating how many of the results are correct.
PredictionThe inferred output of a trained model provided with an input instance. 
PreprocessingThe process of transforming raw data into a more understandable format. 
Pretrained ModelA model trained for relevant tasks, often used as a starting point for fine-tuning. 
Pretraining:The first phase of training foundation models in unsupervised learning. 
Principal Component AnalysisA process that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of linearly uncorrelated variables called principal components.
PriorThe probability distribution that would represent the preexisting beliefs about a specific quantity before new evidence is considered.
PromptInitial context or instruction for the model.
Prompt ChainingUsing multiple prompts to refine a request. 
Prompt EngineeringDesigning effective user requests for AI models. 
PluginsSoftware components that extend the functionality of language models. 
PEMT (Post Edit Machine Translation)Editing machine-translated documents. 
Pooling (Max Pooling)The process of reducing a matrix generated by a convolutional layer to a smaller matrix. 
Post-ProcessingProcedures that filter noisy or imprecise knowledge derived by an algorithm. 

Q (1) – Quotas: The ever-elusive targets you’re supposed to meet.

  • Q-Learning: A type of reinforcement learning that enables AI models to learn and improve iteratively over time. 

R (17) – Remote Work: Pajamas, coffee, and meetings in bed—what a life.

Random ForestAn ensemble learning method that operates by constructing a multitude of decision trees at training time and outputting a combined version (such as the mean or the mode) of the results of each individual tree.
RecallThe fraction of all relevant samples that are correctly classified as positive. 
Rectified Linear UnitA unit employing the rectifier function as an activation function.
Regression (Linear Regression, Logistic Regression)A set of statistical processes for estimating the relationships among variables. 
RegressorA feature, an explanatory variable used as an input to a model. 
RegularizationThe process of introducing additional information in order to prevent overfitting. 
Reinforcement LearningThe subfield of Machine Learning inspired by human behavior studying how an agent should take action in a given environment to maximize some notion of cumulative reward.
Reinforcement Learning from Human Feedback (RLHF)RLHF trains models directly from human feedback, as opposed to from a coded reward stimulus. Humans may score a chatbot’s output and feed those scores back into the model.
Recurrent Neural Networks (RNN)Neural network models used in NLP and speech recognition. 
RelationsIdentifying relationships between elements in statements.
Responsible AIEnsuring transparent, explainable, fair, and sustainable AI. 
Retrieval-Augmented Generation (RAG)Technique for adding external data to AI responses. 
Recommendation EngineA recommendation engine is an AI algorithm that is used to serve users content based on their preferences. Social sites, such as TikTok, and streaming platforms, such as Spotify and YouTube, use recommendation engines to personalize user feeds.
ReproducibilityA methodological crisis in science in which scholars have found that the results of many scientific studies are difficult or impossible to replicate or reproduce on subsequent investigation, either by independent researchers or by the original researchers themselves.
Restricted Boltzmann MachinesA restricted Boltzmann machine (RBM) is a generative stochastic artificial neural network that can learn a probability distribution over its set of inputs. 
Robotic Process Automation (RPA)A type of business operation enhancement utilizing software robots (bots) or AI-driven tools. It’s well known for its role in Conversational Automation, or workflow Automation. Sometimes, it’s termed software robotics, distinct from robot software.
Rules-based Machine Translation (RBMT)Machine translation based on linguistic information. 

S (17) – Self-Motivation: That thing you need to muster before your coffee kicks in.

  • SAO (Subject-Action-Object): Identifying logical functions in sentences.
  • Semantic Network: A knowledge representation system connecting concepts. 

  • Semantic Search: Using natural language technologies to improve user search capabilities. 

  • Semantics: The study of word and sentence meanings. 

  • Semi-Structured Data: Data that doesn’t adhere to a rigid structured format. 

  • Semi-Supervised Learning: A class of supervised learning techniques that also leverages available unlabeled data for training, typically using a small number of labeled instances in combination with a larger amount of unlabeled rows. See also Supervised Learning and Unsupervised Learning. 

  • Sentiment: The general disposition expressed in text. 

  • Sentiment Analysis: The use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affected states and subjective information. 

  • Similarity (and Correlation): Retrieving documents similar to a given one. 

  • Simple Knowledge Organization System (SKOS): A data model for knowledge organization systems. 

  • Slop: Slop is to GenAI what spam is for email. Unwanted AI-generated content.

  • Speech Recognition: Converts spoken language into text using AI. 

  • Statistical Distribution: In statistics, an empirical distribution function is the distribution function associated with the empirical measure of a sample. This cumulative distribution function is a step function that jumps up by 1/n at each of the n data points. Its value at any specified value of the measured variable is the fraction of observations of the measured variable that are less than or equal to the specified value.
  • Structured Data: Data conforming to specific data models and structures. 

  • Supervised Learning: The Machine Learning task of learning a function mapping an input to an output based on example input-output pairs. 

  • Support Vector Machines (SVM): A class of discriminative classifiers formally defined by a separating hyperplane, where for each provided labeled training data point, the algorithm outputs an optimal hyperplane which categorizes new examples. 

  • Synthetic Data: Data generated artificially when real data cannot be collected in sufficient amounts, or when original data doesn’t meet certain requirements. 

  • Syntax: The arrangement of words and phrases to create meaning in language. 

T (16) – “This Meeting Could Have Been an Email”: The universal office truth.

TaggingSee Parts-of-Speech Tagging (POS Tagging).
TaxonomyA predetermined group of classes with hierarchical dependencies.
TemperatureA parameter controlling the randomness of LLM output.
Test SetA collection of sample documents used to measure the accuracy of an ML system. 
Text AnalyticsTechniques to process unstructured text for insights and patterns.
Text SummarizationTechniques that produce short summaries of longer texts.
TensorFlowAn open-source library, popular among the Machine Learning community, for data flow programming across a range of tasks. It is a symbolic math library and is also used for machine learning applications such as neural networks. 
Time Series (Time Series Data)A sequence of data points recorded at specific times and indexed accordingly to their order of occurrence.
Testing (Testing Data)In the context of Supervised Machine Learning, the process of assessing the final performance of a model using hold-out data.
Topic ModelingA category of Unsupervised Machine Learning algorithms that uses clustering to find hidden structures in textual data, and interpret them as topics.
Training DataIn the context of Supervised Machine Learning, the construction of algorithms that can learn from and make predictions from data.
Transfer LearningAn area of Machine Learning that focuses on using knowledge gained to solve a specific problem and apply this knowledge to a different but related problem. 
TransformerA transformer is a type of deep neural network architecture that is made up of multiple encoder and decoder components that are combined in such a way to enable the processing of sequential data such as natural language and time series.
Turing TestA test developed by Alan Turing to evaluate a machine’s ability to exhibit intelligent behavior equivalent to that of a human. The test consists in having the machine chat with a human. If a human evaluator witnessing the conversation from outside the room where the test takes place can’t reliably tell the machine from the human apart, the machine is said to have passed the Turing test.
TokenA token is the basic unit of text that an LLM uses to understand and generate language. It may be a word or parts of a word. Paid LLMs, such as GPT-4’s API, charge users by token
Technological SsingularityThe singularity describes a point in the future where advanced AI becomes more intelligent than humans and technological growth becomes uncontrollable. 

U (5) – Upskilling: The constant chase to stay ahead of AI.

  • Uncertainty: A range of values likely to enclose the true value.
  • Underfitting: A modeling error in statistics and machine learning when a statistical model or machine learning algorithm cannot adequately capture the underlying structure of the data. 

  • Universal Bot: An advanced AI-powered chatbot designed to interact across multiple channels and platforms. It’s capable of understanding and responding to user queries in a consistent, coherent manner, providing a unified and seamless customer experience regardless of the communication medium used. 

  • Unstructured Data: Raw, unprocessed data. Textual data is a perfect example of unstructured data because it is not formatted into specific features. 

  • Unsupervised Learning: Training an algorithm on unlabeled data. 

V (3) – Virtual Assistant: AI, doing its best to replace you one task at a time.

Vanishing / Exploding GradientsA dreaded difficulty and major obstacle to recurrent net performance that data scientists face when training Artificial Neural Networks with gradient-based learning methods and backpropagation, due to the neural network’s weights receiving an update proportional to the partial derivative of the error function with respect to the current weight in each iteration of training.
VarianceAn error due to sensitivity to small fluctuations in the training set computed as the expectation of the squared deviation of a random variable from its mean.
Validation DataA subset of the dataset used in machine learning to tune model hyperparameters.

W (1) – Work-Life Balance: That mythical concept everyone talks about but nobody achieves.

  • Workflow Automation: The process of using software or tools to automate manual tasks or workflows in a business environment. This advanced form of automation learns and improves over time, reducing human intervention, minimizing errors, and enhancing efficiency and productivity. 

X (1) – Excel: The universal tool for crunching numbers, making graphs, and pretending to work.

XAI (Explainable AI)A subfield of AI focused on creating transparent models with clear and understandable explanations. 

Y (4) – “You’re On Mute”: The phrase of 2020 that defines virtual meetings.

  • YYAML (YAML Ain’t Markup Language): YAML is a human-readable data serialization format commonly used for configuration files and data exchange between languages with different data structures. It’s often used in AI and machine learning projects for specifying parameters and configurations. 

  • Yield: While not a specific AI term, “yield” can be relevant in the context of programming and generators. It’s used to generate values or data in a memory-efficient manner, which can be useful in AI applications when dealing with large datasets or generating sequences. 

  • Yottabyte (YB): A yottabyte is a unit of digital information storage that represents one trillion gigabytes (or 2^80 bytes). In the context of AI, it can be relevant when discussing the storage and processing of large datasets. 

  • YUV Color Space: YUV is a color space used in image and video processing. It separates the luminance (Y) and chrominance (U and V) components of an image, which can be useful in various AI applications, including image and video analysis. 

Z (1) – Zero Inbox: That rare state of email nirvana that few ever attain.

Zero-Shot LearningA type of machine learning where the model makes predictions for conditions not seen during training without any fine-tuning.

About Author: Mindy Honcoop

Sources:,, Coursera, NY Times, CNet, Appen, Tech Target, Moveworks, Aisera, Intercom, and HR Grapevine