Aritificial Intelligence (AI): At a fundamental level, AI systems work by processing massive amounts of data through complex algorithms (rules) and specialized models to achieve goals like:
- Learning from patterns (Machine Learning).
- Problem-solving and decision-making.
- Understanding and generating human language (Natural Language Processing).
- Recognizing images and complex data structures.
Agentic AI: A type of artificial intelligence capable of acting on behalf of a user or function to complete tasks or achieve goals. These systems, known as AI agents, often operate with autonomy and limited human intervention, generating responses, making decisions, or interacting with other software tools based on user prompts, instructions, or learned behavior.
Bias: Refers to flaws in the data used to train AI, or a flaw in the design of the algorithm itself, that can amplify existing societal biases.
Chatbot: An AI tool designed to simulate human conversation.
Deep Learning (DL): A method of training an AI system that processes data in a way inspired by the human brain. DL models are typically used to recognize complex patterns in pictures, text, sounds, and other data to produce accurate insights and predictions.
Generative AI (GenAI): A type of artificial intelligence that can create new content (e.g., text, images, audio, or computer code), rather than just analyzing or classifying existing data. The generated output is highly dependent on the training data. A key drawback is that when untrained on a topic, it may create fictional content (hallucinate) or if trained on biased data, the results will also be biased.
Hallucinations: A condition where a Large Language Model (LLM) generates prompt results on items it was not trained on, often creating nonsensical or inaccurate outputs.
Human Oversight: In the context of AI, this ensures that human beings maintain appropriate control, accountability, and ethical responsibility over the design, development, deployment, and operation of Artificial Intelligence systems.
Large Language Model (LLM): A computational model recognized for its ability to achieve general-purpose language generation and other natural language processing tasks, such as classification.
Machine Learning (ML): A branch of AI and computer science that focuses on using data and algorithms to enable AI to imitate the way that humans learn, gradually improving its accuracy.
Prompt Injection: A specialized type of cyber-attack against LLMs, where malicious inputs are disguised as legitimate ones, resulting in the return of erroneous results or the leaking of sensitive information.
Training Data: The data used for the purpose of training AI tools to make decisions, make predictions, or generate content.