AI Glossary

The capability of machines to imitate human intelligence to learn, reason, and solve problems.

Artificial Intelligence (AI) is the ability of computer systems to perform tasks that typically require human intelligence. It can be used across a wide range of applications such as visual perception, speech recognition, decision-making, forecasting, and generating creative content.

A subfield of AI that enables computers to learn patterns from data without being explicitly programmed.

Machine learning is a technology where algorithms are trained using historical data and adapt to new situations by generalizing from that data. For example, systems that sort your emails into spam and non-spam use machine learning.

An advanced machine learning technique that uses multi-layered artificial neural networks inspired by the human brain.

Deep learning relies on deep neural networks capable of automatically extracting complex features from massive datasets (big data). This technology lies at the heart of autonomous vehicles, facial recognition systems, and advanced language models (like ChatGPT).

Massive AI systems capable of understanding and generating human language by reading billions of words.

Large Language Models (LLMs) are AI models trained on colossal text datasets. They can write text, translate, generate code, and provide logical answers to questions. OpenAI's GPT-4 or Google's Gemini models are prime examples of LLMs.

AI systems capable of creating entirely new and original content such as text, images, audio, or video.

Generative AI refers to systems that do not just analyze data but create content from scratch. Generating images with Midjourney, making music with Suno, or having ChatGPT write an article are common applications of Generative AI technology.

The art of optimizing the commands (prompts) given to an AI to get the best and most accurate results.

AI models react based on what you tell them. Prompt engineering is the process of strategically designing commands so the model correctly understands the context, knows its boundaries, and outputs exactly in the desired format.

When an AI confidently provides completely false or fabricated information as if it were true.

LLMs can sometimes produce answers that seem logical but do not align with reality when they encounter unfamiliar topics. This is called 'hallucination'. Because the model statistically predicts words side-by-side, it can form grammatically correct but entirely fictional sentences.

A method that allows an AI to consult reliable external data sources before generating an answer.

RAG is used to prevent language models from hallucinating and to keep them up to date. Instead of searching for the answer solely within its training data, the model first queries a reliable database (e.g., your company's documents) and formulates a response using the retrieved information.

The branch of AI that empowers computers to understand, interpret, and generate human language.

NLP translates everyday human speech or text into mathematical data that computers can process. Voice assistants (Siri, Alexa), automated translation tools, and sentiment analysis software all rely heavily on NLP.

Algorithm networks designed to process data, inspired by the way the human brain works.

Artificial neural networks consist of nodes (neurons) and connections between them. As input data passes through layers, it is processed, and different features of the data are learned at each layer. They are highly successful in tasks like image recognition and audio processing.