Forge of Agents
AI Vocabulary
Essential terms and concepts for AI transformation
Predictive Analytics
Analytics using historical data and patterns to make predictions in given time frames.
Big Data
Very large and complex datasets which are beyond the capabilities of traditional applications for processing data.
Feature Extraction
The process of transforming raw data into numerical features that can be processed while retaining relevant information.
Overfitting
Occurs when a machine learning model learns the details in the training data to the extent that it performs well on this data but poorly on new, unseen data.
Machine Learning
A subset of AI where systems can make predictions or decisions based on data and developing algorithms. ML learns from the data rather than being programmed.
Chatbot
A portmanteau of 'Chat and Robot'. An application which offers a 2-way dialogue with a customer or service user mimicking human interaction using either voice or text.
Training Data
Examples of information given to an AI system which enable it to learn, find patterns, and create new content. For example, historical transaction data including flags denoting confirmed money laundering or fraudulent use of card details.
Ethics of AI
Considerations of potential adverse impact of AI on stakeholders. Current ethical concerns include biases in algorithmic decision-making, privacy violations, and worries over job displacement.
Hallucination
When an AI system provides information which is proven to be false or incorrect in a response and presents it as factual information.
Deep Learning
A function of AI that imitates the human brain by learning from how it makes decisions based on the structuring and processing of information. Examples of unstructured data are images and speech.
Reinforcement Learning
A type of machine learning in which an algorithm is set a goal without specific metrics. The model tests different scenarios and human feedback nudges the model to adjust subsequent scenarios to improve results.
Generative AI
Technology that uses AI to create new content, including text, images, and videos, by learning patterns from existing data. Widely used in applications such as customer service chatbots and content creation tools.
Neural Network
A deep learning model designed to mimic the human brain's structure and function. Neural networks consist of layers of interconnected nodes (neurons) that process data in complex ways to identify patterns and make decisions.
Bias
Systematic errors in AI that result in unfair outcomes, such as disproportionate impact on certain demographic groups. Bias can arise from skewed training data, underrepresentation of certain groups, or developer bias.
Computer Vision
Identifies and analyses objects, people, text, scenes, and activities in images. This technology leverages machine learning algorithms to interpret visual data from photos or videos.
Unsupervised Learning
A type of machine learning that deals with data with no labels and seeks to find the underlying structure in the data.
Artificial Intelligence (AI)
The simulation of human intelligence processes by machines or computer systems. AI can mimic human capabilities with a far more refined and wide-reaching range of responses than historical rules or logic-based software setups.
Large Language Model (LLM)
An AI model trained on extensive text datasets to understand and generate human-like language. LLMs can perform tasks such as text generation, translation, and summarization.
Supervised Learning
A machine learning approach where the model is trained using labelled data, meaning the input data is paired with the correct output. Supervised learning algorithms learn to map inputs to outputs based on this training data.
Natural Language Processing (NLP)
Enables computers to understand, interpret, and respond to human language, both written and spoken. NLP powers features like text analysis, speech recognition, and language translation.
Algorithm
A sequence of rules or instructions given to an AI system to perform a task or solve a problem. Common algorithms include classification, clustering, and regression.
Sentiment Analysis
Using AI to analyse the tone and opinions in a given text. This could be used to denote bias or if the interaction could be misleading.
Guardrails
Restrictions and rules placed on AI systems which ensure data is handled appropriately. For example, they ensure AI systems do not generate offensive or inappropriate content/responses.
Data Mining
The process of sorting through and conducting deep analysis of large data sets to identify patterns which might improve models, uncover new potential opportunities or solve challenges.
Cognitive Computing
The use of computerised models to simulate human thought processes in complex situations where the answers are ambiguous. An alternative way of saying Artificial Intelligence (AI).