Marketing & Advertising Glossary

Embeddings are numerical representations of words, images, or concepts that AI systems use to understand meaning and relationships.

Explainable AI makes machine learning decisions transparent and understandable to humans, crucial for media buying trust and compliance.

Few-shot learning enables AI models to learn from just a handful of examples rather than massive datasets, making rapid adaptation possible in advertising.

Fine-tuning is the process of adjusting a pre-trained AI model with new data to improve its performance for specific advertising tasks.

A large-scale AI model trained on vast amounts of data, capable of performing multiple tasks without task-specific retraining.

Function calling enables AI models to interact with external systems by requesting specific actions or data retrieval through structured outputs.

A specialized processor designed to handle parallel computations, essential for training AI models and processing large datasets in advertising.

Google Gemini is an advanced AI model that can process text, images, video, and audio to help with content creation, analysis, and advertising optimization.

Hugging Face is an open-source AI platform providing pre-built language models and tools for natural language processing tasks in marketing applications.

Jasper AI is an artificial intelligence writing tool that generates marketing copy, social media content, and ad text at scale using natural language processing.

LLAMA is Meta's open-source large language model family designed for natural language processing tasks in AI applications.

An AI system trained on vast amounts of text data to understand and generate human language, powering tools like ChatGPT and marketing automation.

Latency is the delay between a request and response in ad tech systems, critical for real-time bidding and user experience.

A high-dimensional mathematical space where AI models compress and represent complex data patterns in simplified, learnable forms.

LoRA is a technique for efficiently fine-tuning large AI models by training only small adapter layers instead of all parameters.

A mathematical measure that quantifies how poorly an AI model's predictions differ from actual outcomes, guiding model improvement.

Marketing automation uses software to automatically manage repetitive marketing tasks, nurturing leads and personalising customer journeys at scale.

Midjourney is an AI tool that generates high-quality images from text descriptions, widely used by marketers for creative asset production.

A neural network architecture that routes different inputs to specialized sub-networks (experts) for improved efficiency and performance.

AI systems that process and understand multiple types of data (text, images, video, audio) simultaneously to make informed decisions.

Natural Language Processing (NLP) is AI technology that enables machines to understand, interpret, and generate human language for targeted advertising and customer insights.

A neural network is a machine learning system inspired by the human brain that learns patterns from data to make predictions and decisions.

Open-source AI refers to artificial intelligence models and tools with publicly available code that anyone can use, modify, and distribute freely.

Overfitting occurs when an AI model learns training data too well, including noise, reducing its ability to perform on new, unseen data.

Parameters are the internal variables in AI models that get adjusted during training to improve predictions and performance in advertising campaigns.

Performance Max is Google's AI-driven campaign type that automatically optimizes ads across all Google channels using machine learning.

Using historical data and machine learning to forecast future customer behaviour, campaign performance, and market trends.

The practice of crafting precise instructions to AI tools to generate desired outputs for marketing and ad creation tasks.

A security vulnerability where malicious instructions bypass AI safeguards by manipulating how language models interpret user inputs.

Quantization is an AI optimization technique that reduces the precision of model weights to decrease file size and improve inference speed.