Glossary

Adversarial Inputs: Carefully designed inputs that cause AI models to behave unpredictably or generate harmful responses. Reference
Artificial Intelligence (AI): The broader field focused on creating systems capable of tasks requiring human-like intelligence. Reference
Attention Mechanisms: Techniques that allow models to focus on relevant parts of an input sequence. Reference
Autoregressive Text Generation: The process of generating text token by token based on context. Reference

Bias: Systematic errors in AI outputs due to biases present in training data. Reference

Cloud-Based Deployment: Deploying AI models on cloud platforms for scalability and flexibility. Reference
Context Windows: The capacity of a model to process and remember information within a given input sequence. Reference

Data Poisoning: The act of compromising training datasets to introduce harmful behaviors or inaccuracies in AI models. Reference
Deep Learning (DL): A branch of ML that uses multi-layered neural networks to model complex patterns. Reference

Edge Deployment: Running AI models locally on devices for ultra-low latency and offline functionality. Reference
EOS token: An end-of-sequence token is a special marker that signals to the model that it should stop generating further tokens. Reference

Fine-Tuning: The process of adapting a pre-trained model to a specific task or domain through additional training on smaller, task-specific datasets. Reference
Foundation Models: Large-scale, pre-trained models designed to handle a wide range of tasks. Reference

Generative AI (GenAI): AI systems designed to create new content—text, images, or audio—based on learned patterns. Reference

Hallucinations: AI-generated content that appears convincing but has no basis in reality or training data. Reference
Hosted Deployment: Accessing AI models via APIs hosted by external providers. Reference
Hybrid Deployment: Combining cloud and edge capabilities for balanced performance and cost. Reference

Incident Response Plans: Plans designed to address and manage the aftermath of a security breach or cyberattack. Reference
Inference: The application phase, where they generate outputs based on learned patterns without further adjustments to their parameters. Reference

Large Language Models (LLMs): Specialized deep learning models trained on extensive text corpora for language-related tasks like generation and comprehension. Reference

Machine Learning (ML): A subset of AI where systems learn patterns from data rather than being explicitly programmed. Reference
Moderation Endpoints: External tools for assessing and managing content dynamically. Reference
Multi-Head Attention: A mechanism that enables models to analyze multiple aspects of context simultaneously. Reference

Neural Networks: A type of machine learning model inspired by the human brain, consisting of layers of interconnected nodes. Reference

On-Premises Deployment: Hosting AI models internally within an organization’s infrastructure. Reference

Parameters: The internal values that a model learns during training, acting as “weights” that determine the importance of different patterns in input data. Reference
Prompt Engineering: The process of designing and refining prompts to elicit desired responses from AI models. Reference
Prompt Injection: A type of attack where malicious inputs are crafted to manipulate model outputs. Reference

Refusal Pathways: Mechanisms designed to restrict the generation of harmful or undesirable outputs. Reference
Rule-Based Systems: Early AI systems that relied on predefined rules to make decisions. Reference

Specialized Models: AI systems designed to excel at specific tasks or domains. Reference
Serialization: The process of converting a model into a format that can be saved to disk and later loaded into memory for inference. Reference

Token: The smallest unit of text that can be processed by an LLM. Reference
Threat Modeling: The process of identifying and assessing potential threats to a system. Reference
Training: The process where models learn from data through optimization techniques like backpropagation. Reference
Transformers: A type of neural network architecture that uses attention mechanisms to process text data. Reference