Topic 1: Large Language Models (LLMs) – Quick overview

Large Language Models (LLMs) - Quick overview

Large Language Models (LLMs) – Quick overview

Large Language Models, or LLMs, are AI models designed to understand and generate language. They power many of the AI experiences professionals now use for writing, summarising, analysing information, exploring ideas and working with large amounts of text.

The important thing to understand is that an LLM doesn’t retrieve a ready-made answer from a database. It generates a response based on patterns it has learned and the information you provide.

What is a Large Language Model?

A Large Language Model is an AI model trained on very large amounts of text.

During training, the model learns patterns and relationships in language: such as how words, sentences, ideas and concepts commonly relate to one another.

This allows an LLM to respond to instructions and generate language that is relevant to the conversation or task.

What can LLMs do?

LLMs can support many everyday knowledge-work activities, including:

  • Write: create drafts, documents and other written content
  • Summarise: turn large amounts of information into shorter summaries
  • Analyse: identify themes, patterns and relationships in information
  • Explain: make complex subjects easier to understand
  • Transform: rewrite, restructure or reformat existing information
  • Generate ideas: suggest options, alternatives and different approaches

For Product Managers and Business Analysts, these capabilities can be useful across many activities; from exploring a problem to analysing information and preparing work outputs.

How Does an LLM Work? Simple View

At a very high level, you can think about an LLM in five steps:

1. Large amounts of text → 2. Model learns language patterns → 3. You provide an input → 4. Model predicts what should come next → 5. Response is generated

This process happens repeatedly as the model generates its response.

That is why the input and context you provide matter. They help shape what the model generates.

Keep in Mind

LLMs can produce responses that sound extremely confident and convincing, but this doesn’t mean the model thinks, understands or knows things in the same way a human does.

An LLM generates responses based on learned patterns and the context available to it. As a result, an output can sometimes be:

  • Incorrect
  • Incomplete
  • Based on an assumption
  • Unsupported by reliable evidence

Important information should therefore still be reviewed and validated by a human before it is used.

Key Takeaway

LLMs are powerful language engines you provide the task and context, the model generates a response, and you decide what is useful.

Shruti DSM