AI / artificial intelligence terminology

Glossary with example sentences

What is AI / artificial intelligence?

artificial intelligence, AI (noun): the capability of a machine to imitate intelligent human behavior, or the theory and development of computer systems able to perform tasks that typically require human intelligence

Glossary

Each of the following terms, listed in alphabetical order, has 1) a basic definition and 2) an example sentence showing how the term may be used in context.

adversarial attack (noun): manipulating AI's input to get a desired output — Security experts are concerned about adversarial attacks on AI systems.

algorithm (noun): a set of rules or processes followed by a computer — The algorithm efficiently sorted the data.

AI (abbreviation): artificial intelligence
artificial intelligence (noun): the capability of a machine to imitate intelligent human behavior; the theory and development of computer systems able to perform tasks that typically require human intelligence. Also known as AI, "machine intelligence", "computational intelligence", "synthetic intelligence" — Artificial intelligence is transforming the way we live and work.

artificial general intelligence (noun): see strong AI

autonomous (adjective): operating without human intervention — Autonomous vehicles rely heavily on AI.

backpropagation (noun): a method for training neural networks by adjusting weights — Backpropagation helps minimize prediction errors.

bias (noun): unintended preferences in AI decision-making — It's essential to address AI bias to ensure fairness.

chatbot (noun): software that simulates human conversation — Many companies use chatbots for customer support.

cognitive computing (noun): simulating human thought processes in machines — Cognitive computing aims to make AI more human-like.

computer vision (noun): AI's ability to interpret visual data — Computer vision is used in facial recognition systems.

data mining (noun): extracting patterns from large datasets — Data mining helps businesses make informed decisions.

deep learning (noun): a type of machine learning using neural networks with many layers — Deep learning excels in image recognition tasks.

expert system (noun): a computer system that emulates human decision-making — The expert system assists doctors in diagnosis.

GAN (abbreviation): generative adversarial network
generative adversarial network (noun): a model with two neural networks competing — GANs are used to generate realistic images.

generative AI (noun): a subset of artificial intelligence focused on creating new content, such as images, music or text, by training on existing datasets. It's distinct from general AI in that while AI can analyze and process data, generative AI can produce new, original content based on its training. Also known as "generative machine learning", "creative AI" — Generative AI is behind the creation of realistic art and music compositions.

inference (noun): the process of making predictions using a trained model — The system's inference capabilities were impressive.

labeled data (noun): digital text, images etc that have been annotated with predefined tags (such as quantity, spam or not spam, apple or banana) that provide a guide for machine learning algorithms to learn from — We use labeled data to train our NLP models to make predictions or understand speech.

machine learning (noun): a subset of AI where systems learn from data — Machine learning models improve with more data.

narrow AI (noun): see weak AI

NLP (abbreviation): natural language processing (not to be confused with NLP for neuro-linguistic programming)
natural language processing (noun): AI's ability to understand and generate human language — NLP powers chatbots and translation services.

neural network (noun): a computational model inspired by the human brain's interconnected neurons — Neural networks are essential for deep learning tasks.

reinforcement learning (noun): a learning method where the system being trained is rewarded for desired behaviours and punished for undesired behaviours — In a way, reinforcement learning allows a system to learn through trial and error.

robotics (noun): the field of creating and using robots — AI advancements have greatly impacted robotics.

sentiment analysis (noun): determining whether the emotional tone of a digital text is positive, negative or neutral — Companies have large volumes of text data (emails, customer chats, social media etc) and sentiment analysis lets them better gauge customer feedback.

specialized AI (noun): see weak AI

strong AI (noun): a system that can solve problems it’s never been trained to work on. Opposite: weak AI. Also known as "artificial general intelligence" — Strong AI doesn't exist yet and is seen only in Hollywood movies.

supervised learning (noun): machine learning where models are trained using labeled data — Supervised learning requires extensive data annotation.

swarm intelligence (noun): collective behavior of decentralized systems, often inspired by nature — Swarm intelligence is used in drone coordination.

training data (noun): data used to train machine learning models — Accurate training data is crucial for model performance.

Turing test (noun): a test to determine if a machine exhibits human-like intelligence — Few AI systems have passed the Turing test.

unsupervised learning (noun): machine learning without using labeled data — Unsupervised learning discovers hidden patterns in data.

weak AI (noun): a system that can solve problems it's been trained to work on. Opposite: strong AI. Also known as "narrow AI" or "specialized AI". — Next time you do a Google search or hop on a self-driving car, you're using weak AI.

Reference and further resources