LLM in AI is shortened for "Large Language Model." It is used to refers to a type of artificial intelligence model designed and trained to understand and generate human like responses and language. When completely trained and deployed, they are referred to as “Large Language Models. 

These models are typically built and trained using existing information, data and machine learning techniques, particularly deep learning, and, are trained on vast amounts of text, images and  data.

Key features of LLMs include:

1. Scale - LLMs are characterized by their large number of parameters, parameters being the vast numbers that make up the model. Most times, these numbers run in, the billions or even trillions. It is these large numbers that allows them  to capture complex language patterns, questions and nuances.

2. Training Data - LLMs are trained on extensive sources of data, known as datasets. These may include books, articles, websites, and other text sources, enabling them to generate coherent and contextually relevant responses.

3. Capabilities - The capabilities of any LLMs depends on what it is trained on. Depending on what they are trined on, they can perform a wide range of language tasks, such as translation, summarization, text generation, question answering, and more.

Examples of LLMs include GPT-3, GPT-4  and other models developed by organizations like OpenAI, Google, and Microsoft. These models are used in various applications, including chatbots, virtual assistants, content creation tools, and more.