Artificial Intelligence has taken the entire world by storm since the launch of OpenAI’s ChatGPT in November 2022. Since then, industries have clamoured to embrace the new technology, leveraging Artificial Intelligence across all aspects of business. We’ve compiled a list of essential terms that everyone should know to better understand Artificial Intelligence.
The following list of terms highlights key concepts that are essential for building and expanding your knowledge of AI technologies. With these, you'll be better equipped to confidently explore and implement artificial intelligence within your organization.
A
Application Programming Interface (API)
A set of protocols, tools, and definitions that allow developers to interact with an AI system or service. APIs enable developers to integrate AI capabilities, such as machine learning models, natural language processing, or computer vision, into their applications without needing to develop the underlying algorithms or models themselves.
Artificial Intelligence (AI)
The branch of computer science that focuses on creating systems capable of performing tasks that typically require human intelligence. These tasks include problem-solving, learning, reasoning, perception, and language understanding. AI systems can be rule-based, where they follow predefined instructions, or learning-based, where they adapt and improve from experience. AI can be categorized into narrow AI, which is designed for specific tasks, and general AI, which aims to perform any intellectual task that a human can do.
Attribution
The process of identifying and explaining which inputs or features contributed to a particular output or decision made by a model. Attribution is crucial for understanding how AI systems reach their conclusions, ensuring transparency, and enabling users to trust and validate the results. It can also be used to refine models by focusing on the most relevant factors that influence predictions.
Algorithm
A set of rules or instructions that a computer follows to solve a problem or perform a task. In AI and machine learning, algorithms are used to process data, recognize patterns, and make decisions based on input data. Algorithms form the foundation of all AI models, guiding how they learn from data and make predictions.
Annotation
The process of labeling or tagging data, such as images, text, or audio, to provide context or information that can be used to train AI models. Annotations help AI systems understand and learn from data by providing examples of what certain features or patterns represent.
Automation
The use of technology to perform tasks without human intervention, ranging from simple repetitive actions to complex processes, often enhancing efficiency and accuracy.
B
Back-end
The part of a software system that handles the server-side operations, including databases, application logic, and APIs. In the context of AI, the back-end manages data processing, model training, and the execution of algorithms, often hidden from the user but crucial for the system's functionality.
Bard
A language model developed by Google, similar to other AI models like GPT, designed to generate human-like text based on prompts. It is part of the broader category of generative AI, used in applications like chatbots, content creation, and automated responses.
Blackbox
A term used to describe a model or system whose internal workings are not transparent or easily understood. In the context of AI, users can see the inputs and outputs, but the process by which the model arrives at its conclusions is hidden or too complex to interpret, making it challenging to explain or trust the results.
Blocking
The halting of processes or actions based on certain conditions or rules. It can involve preventing specific data, inputs, or operations from being processed, often used for security, data integrity, or to manage system resources.
Bias
The systematic errors in a model’s predictions or decisions that result from prejudiced assumptions, incomplete data, or skewed training processes. Bias can lead to unfair or inaccurate outcomes, often disproportionately affecting certain groups or individuals, making it a critical issue to address in AI development.
C
Chatbot
An AI-powered program designed to simulate conversation with human users, typically through text or voice interactions. Chatbots are typically used in customer service, information retrieval, and personal assistance, and can range from simple, rule-based systems to advanced models capable of understanding and generating natural language.
ChatGPT
An AI language model developed by OpenAI, designed to generate human-like text based on prompts. It can engage in conversations, answer questions, and assist with a wide range of tasks. ChatGPT is part of the GPT (Generative Pre-trained Transformer) series and is known for its ability to understand context and generate coherent responses across various topics.
Computer Vision
A field of AI that enables machines to interpret and understand visual information from the world, such as images or videos. It involves tasks like object detection, image recognition, and scene understanding, allowing AI systems to analyze and respond to visual data similarly to how humans do.
Copilot
An AI-powered assistant designed to help users with tasks by providing real-time suggestions, code completions, or automated solutions. Often used in software development, it acts as a "co-pilot," guiding and enhancing the user's productivity and efficiency by offering context-aware recommendations and support.
D
DALL-E
An AI model developed by OpenAI that generates images from textual descriptions. By understanding and interpreting language, DALL-E can create detailed and diverse visual representations of concepts, objects, or scenes based on the input provided. It combines language understanding with image generation, showcasing the potential of AI in creative and artistic applications.
Data Augmentation
Techniques used to increase the diversity of data available for training models, often by modifying existing data (e.g., rotating or flipping images) to improve model robustness and performance.
Data Sets
Collections of data used to train, validate, and test AI models. These can include images, text, audio, or any other type of information, organized to provide relevant examples that the model can learn from. Quality and diversity in datasets are crucial for building accurate and reliable AI systems.
Deepfake
AI-generated media, typically videos or images, where a person’s likeness is convincingly altered or replaced with someone else’s. Deepfakes use deep learning techniques, particularly neural networks, to create highly realistic but fabricated content, often raising ethical concerns about misinformation and privacy.
Deep Learning
A subset of machine learning that uses neural networks with many layers (hence "deep") to model and understand complex patterns in large datasets. Deep learning is particularly effective for tasks such as image recognition, natural language processing, and speech recognition, enabling AI systems to achieve high levels of accuracy in these areas.
Data Mining
The process of discovering patterns, correlations, and insights from large datasets using statistical and computational techniques. In AI, data mining is used to extract valuable information that can inform decision-making, model training, and predictive analysis, often leading to new knowledge and better business strategies.
E
Edge AI
AI processing that occurs directly on devices like smartphones or sensors, rather than in a centralized cloud or data center, enabling faster responses and reducing the need for constant internet connectivity.
F
Front-end
The part of a software application that interacts directly with the user, typically involving the user interface (UI) and user experience (UX). In AI, the front-end is where users interact with AI-powered tools or applications, such as chatbots or dashboards, while the back-end handles data processing and model execution.
G
Gemini
Google's AI model that combines language understanding with advanced features for multitasking, often positioned as a competitor to models like GPT-4. Gemini is designed to handle complex queries, provide detailed answers, and integrate seamlessly into various Google services, enhancing user experiences with AI-driven insights and capabilities.
Generative AI
A type of artificial intelligence that creates new content, such as text, images, audio, or video, by learning patterns from existing data. Generative AI models, like GPT for text or DALL-E for images, are capable of producing original outputs that can be used in creative applications, content generation, and more.
Generative Pre-trained Transformer (GPT)
A type of AI model developed by OpenAI that uses the Transformer architecture to generate human-like text. GPT models are "pre-trained" on vast amounts of text data and then "fine-tuned" for specific tasks, such as writing, translation, or conversation. They are called "generative" because they can create new content based on the patterns learned during training.
H
Hallucinations
Hallucinations refer to instances where the model generates information or statements that are incorrect, fabricated, or nonsensical. Despite being plausible-sounding, these outputs are not based on actual data or facts, and they represent a significant challenge in ensuring the reliability and accuracy of AI-generated content.
I
Image Generation
The process of creating new images using AI models, often based on text descriptions or other input data. Techniques like GANs (Generative Adversarial Networks) or models like DALL-E are used for this purpose. Image generation is widely used in creative applications, design, and content creation, enabling the production of realistic or stylized images from scratch.
Input
The data or information provided to an AI system or model to process and analyze. Inputs can be in various forms, such as text, images, audio, or numerical data, and they are essential for the AI to perform tasks, make predictions, or generate outputs based on the learned patterns from the training data.
K
Knowledge Graph
A structured representation of information that shows relationships between entities, often used in AI to enhance search engines, recommendation systems, and natural language processing by providing context and understanding connections between different pieces of data.
L
Large Language Models (LLM)
AI models that are trained on vast amounts of text data to understand and generate human-like language. These models, such as GPT-4 or BERT, have billions of parameters and can perform a wide range of language-related tasks, including translation, summarization, question-answering, and text generation. LLMs are known for their ability to understand context and produce coherent, contextually relevant text, making them powerful tools in natural language processing and AI applications.
Learning Model
Any machine learning or AI model that is trained on data to recognize patterns and make predictions or decisions. Learning models can vary in complexity, from simple linear regression models to complex deep neural networks, and are used in applications ranging from image recognition to natural language processing. The model "learns" by adjusting its parameters based on the training data to improve its accuracy over time.
M
Machine Learning (ML)
A subset of artificial intelligence that focuses on developing algorithms and models that enable computers to learn from and make predictions or decisions based on data. Instead of being explicitly programmed to perform a task, ML models identify patterns and relationships in data to improve their performance over time. Machine learning is used in a wide range of applications, including recommendation systems, image recognition, and predictive analytics.
Multimodal
Refers to AI models or systems that can process and integrate multiple types of data inputs, such as text, images, audio, and video, simultaneously. Multimodal AI is designed to understand and generate outputs that combine these different data types, allowing for more comprehensive and context-aware responses or actions. This capability is particularly useful in applications like virtual assistants, content generation, and complex decision-making tasks.
N
Natural Language Processing
A field of AI that focuses on the interaction between computers and human language. NLP involves enabling machines to understand, interpret, and generate human language in a way that is both meaningful and useful. It powers applications like chatbots, translation services, and sentiment analysis.
Neural Network
A computational model inspired by the structure and function of the human brain, consisting of layers of interconnected nodes (neurons). Neural networks are the foundation of many modern AI systems, particularly in deep learning. Each node processes input data and passes the result to the next layer, allowing the network to learn complex patterns and make predictions. Neural networks are widely used in tasks like image recognition, natural language processing, and speech recognition.
O
OpenAI
An artificial intelligence research organization and company that aims to develop and ensure that AI benefits all of humanity. Known for creating advanced AI models like GPT (Generative Pre-trained Transformer) and DALL-E, OpenAI focuses on building safe and powerful AI technologies, conducting research, and providing AI tools and resources to the public. OpenAI operates with a mission to ensure that artificial general intelligence (AGI) is aligned with human values and interests.
Open Source
A collaborative model of software development where source code is freely available for anyone to use, modify, and share. In AI, open source projects allow developers to contribute to and benefit from a collective pool of resources, leading to faster innovation and more accessible AI technologies. Examples include open-source AI libraries like TensorFlow, PyTorch, and scikit-learn, which have become foundational tools in the AI community.
Output
The result or data produced by an AI model or system after processing the input. In machine learning, the output could be a prediction, classification, generated text, image, or any other type of result that the model is designed to produce. Outputs are crucial for evaluating the performance of AI systems and determining how well they meet the intended goals or tasks.
P
Predictive Modeling
The process of creating, testing, and validating models that use historical data to predict future outcomes or behaviors. Predictive modeling involves applying statistical techniques and machine learning algorithms to identify patterns in data, which are then used to make forecasts or inform decision-making. This approach is widely used in fields like finance, healthcare, marketing, and more, where accurate predictions can lead to better strategic planning and risk management.
Q
Quantum Machine Learning
The intersection of quantum computing and machine learning, exploring how quantum algorithms can improve machine learning tasks.
R
Reinforcement Learning from Human Feedback (RLHF)
A technique where AI models are trained using feedback from humans to guide and refine their behavior. Human evaluators provide ratings or rankings on the model's actions, helping it learn more desirable outcomes and better align with human values.
Robots.txt
A file used by websites to instruct search engine crawlers which pages or sections of the site should or should not be indexed or crawled. This helps manage web traffic and protect certain content from being accessed or shown in search engine results. While not directly an AI concept, it plays a crucial role in managing how AI-powered search engines interact with web content.
S
Scrape
The process of extracting data from websites or other online sources, typically using automated tools or scripts. In AI and data science, scraping is often used to collect large amounts of data for analysis, model training, or research. However, scraping can raise legal and ethical concerns, especially if it involves unauthorized access to data or breaches a website’s terms of service.
Synthetic Data
Artificially generated data created to mimic real-world data. It is used to train AI models when real data is scarce, sensitive, or difficult to obtain. Synthetic data can help improve model performance, enhance privacy, and provide more diverse training examples. It’s often generated using techniques like simulations, generative models, or by modifying existing data.
Supervised Learning
A type of machine learning where a model is trained on labeled data, meaning each input comes with an associated correct output. The model learns to map inputs to outputs by finding patterns in the training data. Once trained, it can predict the output for new, unseen inputs. Supervised learning is commonly used in tasks like classification and regression, such as spam detection or predicting house prices.
T
Tokenization
The process of breaking down text into smaller units, called tokens, which can be words, phrases, or even characters. In AI and NLP, tokenization is a crucial preprocessing step that allows models to process and analyze text more effectively by converting it into a format that the model can understand. Tokens are the building blocks for training language models and performing tasks like text generation and sentiment analysis.
Training/Learning
The phase in machine learning where an AI model learns from a labeled dataset by adjusting its internal parameters to minimize errors and improve accuracy. During training, the model processes input data and compares its predictions to the correct outputs, iteratively refining itself to perform better on future data. This process is essential for enabling the model to generalize and make accurate predictions on new, unseen data.
V
Virtual Assistant
An AI-powered software that can perform tasks or services for an individual, often using natural language processing to interact with users.
W
Web Crawler
An automated program or bot that systematically browses the web to index content for search engines or gather data for analysis. Web crawlers visit websites, follow links, and extract information from pages, enabling search engines like Google to update their databases and provide relevant search results. In AI, crawlers are often used to collect large datasets for training models, such as for natural language processing or web scraping tasks.
Curious to expand your ad tech vocabulary? Check out our Ultimate CTV Glossary
Artificial Intelligence has taken the entire world by storm since the launch of OpenAI’s ChatGPT in November 2022. Since then, industries have clamoured to embrace the new technology, leveraging Artificial Intelligence across all aspects of business. We’ve compiled a list of essential terms that everyone should know to better understand Artificial Intelligence.
The following list of terms highlights key concepts that are essential for building and expanding your knowledge of AI technologies. With these, you'll be better equipped to confidently explore and implement artificial intelligence within your organization.
A
Application Programming Interface (API)
A set of protocols, tools, and definitions that allow developers to interact with an AI system or service. APIs enable developers to integrate AI capabilities, such as machine learning models, natural language processing, or computer vision, into their applications without needing to develop the underlying algorithms or models themselves.
Artificial Intelligence (AI)
The branch of computer science that focuses on creating systems capable of performing tasks that typically require human intelligence. These tasks include problem-solving, learning, reasoning, perception, and language understanding. AI systems can be rule-based, where they follow predefined instructions, or learning-based, where they adapt and improve from experience. AI can be categorized into narrow AI, which is designed for specific tasks, and general AI, which aims to perform any intellectual task that a human can do.
Attribution
The process of identifying and explaining which inputs or features contributed to a particular output or decision made by a model. Attribution is crucial for understanding how AI systems reach their conclusions, ensuring transparency, and enabling users to trust and validate the results. It can also be used to refine models by focusing on the most relevant factors that influence predictions.
Algorithm
A set of rules or instructions that a computer follows to solve a problem or perform a task. In AI and machine learning, algorithms are used to process data, recognize patterns, and make decisions based on input data. Algorithms form the foundation of all AI models, guiding how they learn from data and make predictions.
Annotation
The process of labeling or tagging data, such as images, text, or audio, to provide context or information that can be used to train AI models. Annotations help AI systems understand and learn from data by providing examples of what certain features or patterns represent.
Automation
The use of technology to perform tasks without human intervention, ranging from simple repetitive actions to complex processes, often enhancing efficiency and accuracy.
B
Back-end
The part of a software system that handles the server-side operations, including databases, application logic, and APIs. In the context of AI, the back-end manages data processing, model training, and the execution of algorithms, often hidden from the user but crucial for the system's functionality.
Bard
A language model developed by Google, similar to other AI models like GPT, designed to generate human-like text based on prompts. It is part of the broader category of generative AI, used in applications like chatbots, content creation, and automated responses.
Blackbox
A term used to describe a model or system whose internal workings are not transparent or easily understood. In the context of AI, users can see the inputs and outputs, but the process by which the model arrives at its conclusions is hidden or too complex to interpret, making it challenging to explain or trust the results.
Blocking
The halting of processes or actions based on certain conditions or rules. It can involve preventing specific data, inputs, or operations from being processed, often used for security, data integrity, or to manage system resources.
Bias
The systematic errors in a model’s predictions or decisions that result from prejudiced assumptions, incomplete data, or skewed training processes. Bias can lead to unfair or inaccurate outcomes, often disproportionately affecting certain groups or individuals, making it a critical issue to address in AI development.
C
Chatbot
An AI-powered program designed to simulate conversation with human users, typically through text or voice interactions. Chatbots are typically used in customer service, information retrieval, and personal assistance, and can range from simple, rule-based systems to advanced models capable of understanding and generating natural language.
ChatGPT
An AI language model developed by OpenAI, designed to generate human-like text based on prompts. It can engage in conversations, answer questions, and assist with a wide range of tasks. ChatGPT is part of the GPT (Generative Pre-trained Transformer) series and is known for its ability to understand context and generate coherent responses across various topics.
Computer Vision
A field of AI that enables machines to interpret and understand visual information from the world, such as images or videos. It involves tasks like object detection, image recognition, and scene understanding, allowing AI systems to analyze and respond to visual data similarly to how humans do.
Copilot
An AI-powered assistant designed to help users with tasks by providing real-time suggestions, code completions, or automated solutions. Often used in software development, it acts as a "co-pilot," guiding and enhancing the user's productivity and efficiency by offering context-aware recommendations and support.
D
DALL-E
An AI model developed by OpenAI that generates images from textual descriptions. By understanding and interpreting language, DALL-E can create detailed and diverse visual representations of concepts, objects, or scenes based on the input provided. It combines language understanding with image generation, showcasing the potential of AI in creative and artistic applications.
Data Augmentation
Techniques used to increase the diversity of data available for training models, often by modifying existing data (e.g., rotating or flipping images) to improve model robustness and performance.
Data Sets
Collections of data used to train, validate, and test AI models. These can include images, text, audio, or any other type of information, organized to provide relevant examples that the model can learn from. Quality and diversity in datasets are crucial for building accurate and reliable AI systems.
Deepfake
AI-generated media, typically videos or images, where a person’s likeness is convincingly altered or replaced with someone else’s. Deepfakes use deep learning techniques, particularly neural networks, to create highly realistic but fabricated content, often raising ethical concerns about misinformation and privacy.
Deep Learning
A subset of machine learning that uses neural networks with many layers (hence "deep") to model and understand complex patterns in large datasets. Deep learning is particularly effective for tasks such as image recognition, natural language processing, and speech recognition, enabling AI systems to achieve high levels of accuracy in these areas.
Data Mining
The process of discovering patterns, correlations, and insights from large datasets using statistical and computational techniques. In AI, data mining is used to extract valuable information that can inform decision-making, model training, and predictive analysis, often leading to new knowledge and better business strategies.
E
Edge AI
AI processing that occurs directly on devices like smartphones or sensors, rather than in a centralized cloud or data center, enabling faster responses and reducing the need for constant internet connectivity.
F
Front-end
The part of a software application that interacts directly with the user, typically involving the user interface (UI) and user experience (UX). In AI, the front-end is where users interact with AI-powered tools or applications, such as chatbots or dashboards, while the back-end handles data processing and model execution.
G
Gemini
Google's AI model that combines language understanding with advanced features for multitasking, often positioned as a competitor to models like GPT-4. Gemini is designed to handle complex queries, provide detailed answers, and integrate seamlessly into various Google services, enhancing user experiences with AI-driven insights and capabilities.
Generative AI
A type of artificial intelligence that creates new content, such as text, images, audio, or video, by learning patterns from existing data. Generative AI models, like GPT for text or DALL-E for images, are capable of producing original outputs that can be used in creative applications, content generation, and more.
Generative Pre-trained Transformer (GPT)
A type of AI model developed by OpenAI that uses the Transformer architecture to generate human-like text. GPT models are "pre-trained" on vast amounts of text data and then "fine-tuned" for specific tasks, such as writing, translation, or conversation. They are called "generative" because they can create new content based on the patterns learned during training.
H
Hallucinations
Hallucinations refer to instances where the model generates information or statements that are incorrect, fabricated, or nonsensical. Despite being plausible-sounding, these outputs are not based on actual data or facts, and they represent a significant challenge in ensuring the reliability and accuracy of AI-generated content.
I
Image Generation
The process of creating new images using AI models, often based on text descriptions or other input data. Techniques like GANs (Generative Adversarial Networks) or models like DALL-E are used for this purpose. Image generation is widely used in creative applications, design, and content creation, enabling the production of realistic or stylized images from scratch.
Input
The data or information provided to an AI system or model to process and analyze. Inputs can be in various forms, such as text, images, audio, or numerical data, and they are essential for the AI to perform tasks, make predictions, or generate outputs based on the learned patterns from the training data.
K
Knowledge Graph
A structured representation of information that shows relationships between entities, often used in AI to enhance search engines, recommendation systems, and natural language processing by providing context and understanding connections between different pieces of data.
L
Large Language Models (LLM)
AI models that are trained on vast amounts of text data to understand and generate human-like language. These models, such as GPT-4 or BERT, have billions of parameters and can perform a wide range of language-related tasks, including translation, summarization, question-answering, and text generation. LLMs are known for their ability to understand context and produce coherent, contextually relevant text, making them powerful tools in natural language processing and AI applications.
Learning Model
Any machine learning or AI model that is trained on data to recognize patterns and make predictions or decisions. Learning models can vary in complexity, from simple linear regression models to complex deep neural networks, and are used in applications ranging from image recognition to natural language processing. The model "learns" by adjusting its parameters based on the training data to improve its accuracy over time.
M
Machine Learning (ML)
A subset of artificial intelligence that focuses on developing algorithms and models that enable computers to learn from and make predictions or decisions based on data. Instead of being explicitly programmed to perform a task, ML models identify patterns and relationships in data to improve their performance over time. Machine learning is used in a wide range of applications, including recommendation systems, image recognition, and predictive analytics.
Multimodal
Refers to AI models or systems that can process and integrate multiple types of data inputs, such as text, images, audio, and video, simultaneously. Multimodal AI is designed to understand and generate outputs that combine these different data types, allowing for more comprehensive and context-aware responses or actions. This capability is particularly useful in applications like virtual assistants, content generation, and complex decision-making tasks.
N
Natural Language Processing
A field of AI that focuses on the interaction between computers and human language. NLP involves enabling machines to understand, interpret, and generate human language in a way that is both meaningful and useful. It powers applications like chatbots, translation services, and sentiment analysis.
Neural Network
A computational model inspired by the structure and function of the human brain, consisting of layers of interconnected nodes (neurons). Neural networks are the foundation of many modern AI systems, particularly in deep learning. Each node processes input data and passes the result to the next layer, allowing the network to learn complex patterns and make predictions. Neural networks are widely used in tasks like image recognition, natural language processing, and speech recognition.
O
OpenAI
An artificial intelligence research organization and company that aims to develop and ensure that AI benefits all of humanity. Known for creating advanced AI models like GPT (Generative Pre-trained Transformer) and DALL-E, OpenAI focuses on building safe and powerful AI technologies, conducting research, and providing AI tools and resources to the public. OpenAI operates with a mission to ensure that artificial general intelligence (AGI) is aligned with human values and interests.
Open Source
A collaborative model of software development where source code is freely available for anyone to use, modify, and share. In AI, open source projects allow developers to contribute to and benefit from a collective pool of resources, leading to faster innovation and more accessible AI technologies. Examples include open-source AI libraries like TensorFlow, PyTorch, and scikit-learn, which have become foundational tools in the AI community.
Output
The result or data produced by an AI model or system after processing the input. In machine learning, the output could be a prediction, classification, generated text, image, or any other type of result that the model is designed to produce. Outputs are crucial for evaluating the performance of AI systems and determining how well they meet the intended goals or tasks.
P
Predictive Modeling
The process of creating, testing, and validating models that use historical data to predict future outcomes or behaviors. Predictive modeling involves applying statistical techniques and machine learning algorithms to identify patterns in data, which are then used to make forecasts or inform decision-making. This approach is widely used in fields like finance, healthcare, marketing, and more, where accurate predictions can lead to better strategic planning and risk management.
Q
Quantum Machine Learning
The intersection of quantum computing and machine learning, exploring how quantum algorithms can improve machine learning tasks.
R
Reinforcement Learning from Human Feedback (RLHF)
A technique where AI models are trained using feedback from humans to guide and refine their behavior. Human evaluators provide ratings or rankings on the model's actions, helping it learn more desirable outcomes and better align with human values.
Robots.txt
A file used by websites to instruct search engine crawlers which pages or sections of the site should or should not be indexed or crawled. This helps manage web traffic and protect certain content from being accessed or shown in search engine results. While not directly an AI concept, it plays a crucial role in managing how AI-powered search engines interact with web content.
S
Scrape
The process of extracting data from websites or other online sources, typically using automated tools or scripts. In AI and data science, scraping is often used to collect large amounts of data for analysis, model training, or research. However, scraping can raise legal and ethical concerns, especially if it involves unauthorized access to data or breaches a website’s terms of service.
Synthetic Data
Artificially generated data created to mimic real-world data. It is used to train AI models when real data is scarce, sensitive, or difficult to obtain. Synthetic data can help improve model performance, enhance privacy, and provide more diverse training examples. It’s often generated using techniques like simulations, generative models, or by modifying existing data.
Supervised Learning
A type of machine learning where a model is trained on labeled data, meaning each input comes with an associated correct output. The model learns to map inputs to outputs by finding patterns in the training data. Once trained, it can predict the output for new, unseen inputs. Supervised learning is commonly used in tasks like classification and regression, such as spam detection or predicting house prices.
T
Tokenization
The process of breaking down text into smaller units, called tokens, which can be words, phrases, or even characters. In AI and NLP, tokenization is a crucial preprocessing step that allows models to process and analyze text more effectively by converting it into a format that the model can understand. Tokens are the building blocks for training language models and performing tasks like text generation and sentiment analysis.
Training/Learning
The phase in machine learning where an AI model learns from a labeled dataset by adjusting its internal parameters to minimize errors and improve accuracy. During training, the model processes input data and compares its predictions to the correct outputs, iteratively refining itself to perform better on future data. This process is essential for enabling the model to generalize and make accurate predictions on new, unseen data.
V
Virtual Assistant
An AI-powered software that can perform tasks or services for an individual, often using natural language processing to interact with users.
W
Web Crawler
An automated program or bot that systematically browses the web to index content for search engines or gather data for analysis. Web crawlers visit websites, follow links, and extract information from pages, enabling search engines like Google to update their databases and provide relevant search results. In AI, crawlers are often used to collect large datasets for training models, such as for natural language processing or web scraping tasks.
Curious to expand your ad tech vocabulary? Check out our Ultimate CTV Glossary
Behind Headlines: 180 Seconds in Ad Tech is a short 3-minute podcast exploring the news in the digital advertising industry. Ad tech is a fast-growing industry with many updates happening daily. As it can be hard for most to keep up with the latest news, the Sharethrough team wanted to create an audio series compiling notable mentions each week.
Artificial Intelligence has taken the entire world by storm since the launch of OpenAI’s ChatGPT in November 2022. Since then, industries have clamoured to embrace the new technology, leveraging Artificial Intelligence across all aspects of business. We’ve compiled a list of essential terms that everyone should know to better understand Artificial Intelligence.
The following list of terms highlights key concepts that are essential for building and expanding your knowledge of AI technologies. With these, you'll be better equipped to confidently explore and implement artificial intelligence within your organization.
A
Application Programming Interface (API)
A set of protocols, tools, and definitions that allow developers to interact with an AI system or service. APIs enable developers to integrate AI capabilities, such as machine learning models, natural language processing, or computer vision, into their applications without needing to develop the underlying algorithms or models themselves.
Artificial Intelligence (AI)
The branch of computer science that focuses on creating systems capable of performing tasks that typically require human intelligence. These tasks include problem-solving, learning, reasoning, perception, and language understanding. AI systems can be rule-based, where they follow predefined instructions, or learning-based, where they adapt and improve from experience. AI can be categorized into narrow AI, which is designed for specific tasks, and general AI, which aims to perform any intellectual task that a human can do.
Attribution
The process of identifying and explaining which inputs or features contributed to a particular output or decision made by a model. Attribution is crucial for understanding how AI systems reach their conclusions, ensuring transparency, and enabling users to trust and validate the results. It can also be used to refine models by focusing on the most relevant factors that influence predictions.
Algorithm
A set of rules or instructions that a computer follows to solve a problem or perform a task. In AI and machine learning, algorithms are used to process data, recognize patterns, and make decisions based on input data. Algorithms form the foundation of all AI models, guiding how they learn from data and make predictions.
Annotation
The process of labeling or tagging data, such as images, text, or audio, to provide context or information that can be used to train AI models. Annotations help AI systems understand and learn from data by providing examples of what certain features or patterns represent.
Automation
The use of technology to perform tasks without human intervention, ranging from simple repetitive actions to complex processes, often enhancing efficiency and accuracy.
B
Back-end
The part of a software system that handles the server-side operations, including databases, application logic, and APIs. In the context of AI, the back-end manages data processing, model training, and the execution of algorithms, often hidden from the user but crucial for the system's functionality.
Bard
A language model developed by Google, similar to other AI models like GPT, designed to generate human-like text based on prompts. It is part of the broader category of generative AI, used in applications like chatbots, content creation, and automated responses.
Blackbox
A term used to describe a model or system whose internal workings are not transparent or easily understood. In the context of AI, users can see the inputs and outputs, but the process by which the model arrives at its conclusions is hidden or too complex to interpret, making it challenging to explain or trust the results.
Blocking
The halting of processes or actions based on certain conditions or rules. It can involve preventing specific data, inputs, or operations from being processed, often used for security, data integrity, or to manage system resources.
Bias
The systematic errors in a model’s predictions or decisions that result from prejudiced assumptions, incomplete data, or skewed training processes. Bias can lead to unfair or inaccurate outcomes, often disproportionately affecting certain groups or individuals, making it a critical issue to address in AI development.
C
Chatbot
An AI-powered program designed to simulate conversation with human users, typically through text or voice interactions. Chatbots are typically used in customer service, information retrieval, and personal assistance, and can range from simple, rule-based systems to advanced models capable of understanding and generating natural language.
ChatGPT
An AI language model developed by OpenAI, designed to generate human-like text based on prompts. It can engage in conversations, answer questions, and assist with a wide range of tasks. ChatGPT is part of the GPT (Generative Pre-trained Transformer) series and is known for its ability to understand context and generate coherent responses across various topics.
Computer Vision
A field of AI that enables machines to interpret and understand visual information from the world, such as images or videos. It involves tasks like object detection, image recognition, and scene understanding, allowing AI systems to analyze and respond to visual data similarly to how humans do.
Copilot
An AI-powered assistant designed to help users with tasks by providing real-time suggestions, code completions, or automated solutions. Often used in software development, it acts as a "co-pilot," guiding and enhancing the user's productivity and efficiency by offering context-aware recommendations and support.
D
DALL-E
An AI model developed by OpenAI that generates images from textual descriptions. By understanding and interpreting language, DALL-E can create detailed and diverse visual representations of concepts, objects, or scenes based on the input provided. It combines language understanding with image generation, showcasing the potential of AI in creative and artistic applications.
Data Augmentation
Techniques used to increase the diversity of data available for training models, often by modifying existing data (e.g., rotating or flipping images) to improve model robustness and performance.
Data Sets
Collections of data used to train, validate, and test AI models. These can include images, text, audio, or any other type of information, organized to provide relevant examples that the model can learn from. Quality and diversity in datasets are crucial for building accurate and reliable AI systems.
Deepfake
AI-generated media, typically videos or images, where a person’s likeness is convincingly altered or replaced with someone else’s. Deepfakes use deep learning techniques, particularly neural networks, to create highly realistic but fabricated content, often raising ethical concerns about misinformation and privacy.
Deep Learning
A subset of machine learning that uses neural networks with many layers (hence "deep") to model and understand complex patterns in large datasets. Deep learning is particularly effective for tasks such as image recognition, natural language processing, and speech recognition, enabling AI systems to achieve high levels of accuracy in these areas.
Data Mining
The process of discovering patterns, correlations, and insights from large datasets using statistical and computational techniques. In AI, data mining is used to extract valuable information that can inform decision-making, model training, and predictive analysis, often leading to new knowledge and better business strategies.
E
Edge AI
AI processing that occurs directly on devices like smartphones or sensors, rather than in a centralized cloud or data center, enabling faster responses and reducing the need for constant internet connectivity.
F
Front-end
The part of a software application that interacts directly with the user, typically involving the user interface (UI) and user experience (UX). In AI, the front-end is where users interact with AI-powered tools or applications, such as chatbots or dashboards, while the back-end handles data processing and model execution.
G
Gemini
Google's AI model that combines language understanding with advanced features for multitasking, often positioned as a competitor to models like GPT-4. Gemini is designed to handle complex queries, provide detailed answers, and integrate seamlessly into various Google services, enhancing user experiences with AI-driven insights and capabilities.
Generative AI
A type of artificial intelligence that creates new content, such as text, images, audio, or video, by learning patterns from existing data. Generative AI models, like GPT for text or DALL-E for images, are capable of producing original outputs that can be used in creative applications, content generation, and more.
Generative Pre-trained Transformer (GPT)
A type of AI model developed by OpenAI that uses the Transformer architecture to generate human-like text. GPT models are "pre-trained" on vast amounts of text data and then "fine-tuned" for specific tasks, such as writing, translation, or conversation. They are called "generative" because they can create new content based on the patterns learned during training.
H
Hallucinations
Hallucinations refer to instances where the model generates information or statements that are incorrect, fabricated, or nonsensical. Despite being plausible-sounding, these outputs are not based on actual data or facts, and they represent a significant challenge in ensuring the reliability and accuracy of AI-generated content.
I
Image Generation
The process of creating new images using AI models, often based on text descriptions or other input data. Techniques like GANs (Generative Adversarial Networks) or models like DALL-E are used for this purpose. Image generation is widely used in creative applications, design, and content creation, enabling the production of realistic or stylized images from scratch.
Input
The data or information provided to an AI system or model to process and analyze. Inputs can be in various forms, such as text, images, audio, or numerical data, and they are essential for the AI to perform tasks, make predictions, or generate outputs based on the learned patterns from the training data.
K
Knowledge Graph
A structured representation of information that shows relationships between entities, often used in AI to enhance search engines, recommendation systems, and natural language processing by providing context and understanding connections between different pieces of data.
L
Large Language Models (LLM)
AI models that are trained on vast amounts of text data to understand and generate human-like language. These models, such as GPT-4 or BERT, have billions of parameters and can perform a wide range of language-related tasks, including translation, summarization, question-answering, and text generation. LLMs are known for their ability to understand context and produce coherent, contextually relevant text, making them powerful tools in natural language processing and AI applications.
Learning Model
Any machine learning or AI model that is trained on data to recognize patterns and make predictions or decisions. Learning models can vary in complexity, from simple linear regression models to complex deep neural networks, and are used in applications ranging from image recognition to natural language processing. The model "learns" by adjusting its parameters based on the training data to improve its accuracy over time.
M
Machine Learning (ML)
A subset of artificial intelligence that focuses on developing algorithms and models that enable computers to learn from and make predictions or decisions based on data. Instead of being explicitly programmed to perform a task, ML models identify patterns and relationships in data to improve their performance over time. Machine learning is used in a wide range of applications, including recommendation systems, image recognition, and predictive analytics.
Multimodal
Refers to AI models or systems that can process and integrate multiple types of data inputs, such as text, images, audio, and video, simultaneously. Multimodal AI is designed to understand and generate outputs that combine these different data types, allowing for more comprehensive and context-aware responses or actions. This capability is particularly useful in applications like virtual assistants, content generation, and complex decision-making tasks.
N
Natural Language Processing
A field of AI that focuses on the interaction between computers and human language. NLP involves enabling machines to understand, interpret, and generate human language in a way that is both meaningful and useful. It powers applications like chatbots, translation services, and sentiment analysis.
Neural Network
A computational model inspired by the structure and function of the human brain, consisting of layers of interconnected nodes (neurons). Neural networks are the foundation of many modern AI systems, particularly in deep learning. Each node processes input data and passes the result to the next layer, allowing the network to learn complex patterns and make predictions. Neural networks are widely used in tasks like image recognition, natural language processing, and speech recognition.
O
OpenAI
An artificial intelligence research organization and company that aims to develop and ensure that AI benefits all of humanity. Known for creating advanced AI models like GPT (Generative Pre-trained Transformer) and DALL-E, OpenAI focuses on building safe and powerful AI technologies, conducting research, and providing AI tools and resources to the public. OpenAI operates with a mission to ensure that artificial general intelligence (AGI) is aligned with human values and interests.
Open Source
A collaborative model of software development where source code is freely available for anyone to use, modify, and share. In AI, open source projects allow developers to contribute to and benefit from a collective pool of resources, leading to faster innovation and more accessible AI technologies. Examples include open-source AI libraries like TensorFlow, PyTorch, and scikit-learn, which have become foundational tools in the AI community.
Output
The result or data produced by an AI model or system after processing the input. In machine learning, the output could be a prediction, classification, generated text, image, or any other type of result that the model is designed to produce. Outputs are crucial for evaluating the performance of AI systems and determining how well they meet the intended goals or tasks.
P
Predictive Modeling
The process of creating, testing, and validating models that use historical data to predict future outcomes or behaviors. Predictive modeling involves applying statistical techniques and machine learning algorithms to identify patterns in data, which are then used to make forecasts or inform decision-making. This approach is widely used in fields like finance, healthcare, marketing, and more, where accurate predictions can lead to better strategic planning and risk management.
Q
Quantum Machine Learning
The intersection of quantum computing and machine learning, exploring how quantum algorithms can improve machine learning tasks.
R
Reinforcement Learning from Human Feedback (RLHF)
A technique where AI models are trained using feedback from humans to guide and refine their behavior. Human evaluators provide ratings or rankings on the model's actions, helping it learn more desirable outcomes and better align with human values.
Robots.txt
A file used by websites to instruct search engine crawlers which pages or sections of the site should or should not be indexed or crawled. This helps manage web traffic and protect certain content from being accessed or shown in search engine results. While not directly an AI concept, it plays a crucial role in managing how AI-powered search engines interact with web content.
S
Scrape
The process of extracting data from websites or other online sources, typically using automated tools or scripts. In AI and data science, scraping is often used to collect large amounts of data for analysis, model training, or research. However, scraping can raise legal and ethical concerns, especially if it involves unauthorized access to data or breaches a website’s terms of service.
Synthetic Data
Artificially generated data created to mimic real-world data. It is used to train AI models when real data is scarce, sensitive, or difficult to obtain. Synthetic data can help improve model performance, enhance privacy, and provide more diverse training examples. It’s often generated using techniques like simulations, generative models, or by modifying existing data.
Supervised Learning
A type of machine learning where a model is trained on labeled data, meaning each input comes with an associated correct output. The model learns to map inputs to outputs by finding patterns in the training data. Once trained, it can predict the output for new, unseen inputs. Supervised learning is commonly used in tasks like classification and regression, such as spam detection or predicting house prices.
T
Tokenization
The process of breaking down text into smaller units, called tokens, which can be words, phrases, or even characters. In AI and NLP, tokenization is a crucial preprocessing step that allows models to process and analyze text more effectively by converting it into a format that the model can understand. Tokens are the building blocks for training language models and performing tasks like text generation and sentiment analysis.
Training/Learning
The phase in machine learning where an AI model learns from a labeled dataset by adjusting its internal parameters to minimize errors and improve accuracy. During training, the model processes input data and compares its predictions to the correct outputs, iteratively refining itself to perform better on future data. This process is essential for enabling the model to generalize and make accurate predictions on new, unseen data.
V
Virtual Assistant
An AI-powered software that can perform tasks or services for an individual, often using natural language processing to interact with users.
W
Web Crawler
An automated program or bot that systematically browses the web to index content for search engines or gather data for analysis. Web crawlers visit websites, follow links, and extract information from pages, enabling search engines like Google to update their databases and provide relevant search results. In AI, crawlers are often used to collect large datasets for training models, such as for natural language processing or web scraping tasks.
Curious to expand your ad tech vocabulary? Check out our Ultimate CTV Glossary
Founded in 2015, Calibrate is a yearly conference for new engineering managers hosted by seasoned engineering managers. The experience level of the speakers ranges from newcomers all the way through senior engineering leaders with over twenty years of experience in the field. Each speaker is greatly concerned about the craft of engineering management. Organized and hosted by Sharethrough, it was conducted yearly in September, from 2015-2019 in San Francisco, California.
Stay Up-to-Date—
Subscribe to our newsletter and receive cutting-edge digital advertising insights, including our weekly Behind Headlines episodes, delivered right to your inbox.