Adverse Impact

Harm caused to stakeholders through misuse, system failures, or unintended outcomes.

Example: Facial recognition misidentifying individuals, leading to reputational or legal consequences.

Algorithm

A set of rules or processes followed in calculations or decision-making by AI systems.

Example: The sequence of steps an AI uses to detect motion in CCTV footage.

Anomaly Detection

Identifying unusual patterns or behaviours that deviate from the norm.

Example: Spotting unusual movements in surveillance footage that could signal a break-in.

Artificial Intelligence (AI)

Computer systems that can perform tasks usually requiring human intelligence.

Example: An AI-powered video analytics system recognising suspicious behaviour.

Autonomous System

A system that can operate and make decisions with minimal or no human input.

Example: A drone carrying out perimeter surveillance automatically.

Big Data

Extremely large datasets that can be analysed computationally to reveal patterns and trends.

Example: Analysing years of access control logs to detect unusual activity.

Bias

Unfair skewing in AI outputs due to flawed training data or processes.

Example: A system misclassifying individuals due to under-representation in its dataset.

Black Box

AI systems where the decision-making process is not easily understood.

Example: A predictive policing tool giving recommendations without clarity on how it reached them.

Chatbot

An AI program that simulates human conversation.

Example: A virtual security assistant answering staff questions via messaging apps.

Computer Vision

AI that allows machines to interpret and understand visual information.

Example: CCTV cameras detecting unattended bags.

Data Labelling

The process of annotating data so it can be used to train AI models.

Example: Tagging CCTV images to train an algorithm to spot vehicles.

Deep Learning

A subset of machine learning using neural networks with many layers.

Example: Facial recognition systems that improve accuracy over time.

Edge Computing

Processing data close to where it’s generated rather than sending it to a central server.

Example: Cameras analysing video on-site rather than uploading all footage to the cloud.

Explainable AI (XAI)

AI designed to make decision-making processes transparent and understandable.

Example: An intrusion detection system showing why it flagged certain behaviour as suspicious.

Facial Recognition

Technology that identifies or verifies a person’s identity using facial features.

Example: Verifying access at secure facilities.

Generative AI

AI capable of creating new content such as text, images, or video.

Example: Simulating potential threat scenarios for training purposes.

Ground Truth

Accurate data used as a benchmark to train and evaluate AI models.

Example: Verified footage of known incidents used to validate video analytics accuracy.

Image Recognition

The ability of AI to identify objects, places, or people in images.

Example: Detecting number plates via ANPR cameras.

Inference

The process of applying a trained AI model to new data to generate predictions or classifications.

Example: An AI identifying whether movement in footage is human or animal.

Internet of Things (IoT)

A network of interconnected devices that collect and exchange data.

Example: Smart locks and sensors feeding data into a central AI platform.

Machine Learning

AI systems that improve performance automatically through experience and data.

Example: An access control system learning regular user behaviour to flag anomalies.

Model Training

The process of teaching an AI system by feeding it large datasets.

Example: Training a system on thousands of hours of surveillance footage.

Natural Language Processing (NLP)

AI that enables computers to understand and process human language.

Example: Analysing incident reports written by staff to detect recurring issues.

Neural Network

Computing systems modelled after the human brain, capable of recognising complex patterns.

Example: Analysing vast amounts of CCTV footage for threat detection.

Object Detection

The process of identifying and locating objects within an image or video.

Example: Detecting unattended luggage in a transport hub.

Overfitting

When an AI model performs well on training data but poorly on new, unseen data.

Example: A surveillance model working only in daylight conditions it was trained on.

Pattern Recognition

The ability of AI to identify recurring structures or sequences.

Example: Recognising repeat visitor behaviours across multiple cameras.

Predictive Analytics

AI techniques used to forecast future events based on data patterns.

Example: Predicting likely times and locations of security breaches.

Privacy by Design

Embedding privacy and data protection considerations into the development of systems from the start.

Example: Designing CCTV systems with built-in anonymisation features.

Reinforcement Learning

A type of machine learning where systems learn by trial and error with feedback from their actions.

Example: A robot patrol learning optimal routes through feedback.

Robotics

The design and use of machines capable of carrying out complex tasks.

Example: Security robots conducting autonomous patrols in warehouses.

Scalability

The ability of a system to handle growing amounts of work or to be expanded.

Example: Expanding an AI surveillance system from one site to multiple facilities.

Sentiment Analysis

Using NLP to determine the emotional tone behind words.