AI

Large Language Model (LLM)

An AI model trained on vast text to understand and generate language.

A large language model is an AI system trained on enormous amounts of text to predict and generate language. LLMs power chatbots, writing assistants, and AI features, responding to natural-language prompts.

Designing products on top of LLMs means accounting for their strengths (fluent, flexible output) and limits (uncertainty, hallucination) through trust-building UI patterns.

Key characteristics

  • A neural network trained on vast text to predict and generate language.
  • Built on the transformer architecture; capabilities scale with data and parameters.
  • Powers chat assistants, writing, coding and summarisation.
  • Limited by knowledge cut-offs and a tendency to hallucinate without grounding.

Example

When you ask a chatbot to rewrite an email more politely, an LLM predicts a fluent, on-tone version word by word from patterns learned in training.

Frequently asked questions

How does a large language model work?

It predicts the next token (word piece) given prior context, having learned statistical patterns of language from massive training data. Generating text is repeated next-token prediction.

Why do LLMs make mistakes?

They generate plausible text from patterns, not verified facts, so they can hallucinate. They also have training cut-off dates. Grounding techniques like RAG reduce — but don't eliminate — errors.

Need this done, not just defined?

I design and build it — Framer sites, product UI/UX, and more.

Get a fixed quote