Glossary

A way for software systems to communicate with each other. AI APIs allow developers to integrate AI capabilities into other applications, including legal technology platforms.

A broad field of computer science focused on building systems capable of performing tasks that typically require human intelligence, such as pattern recognition, language understanding, and decision-making.

Systematic errors in AI outputs that reflect prejudices in training data or model design. In legal contexts, this can affect sentencing recommendations, hiring tools, and risk assessments.

The maximum amount of text (measured in tokens) that an LLM can process in a single interaction. Larger context windows allow processing of longer documents.

A structured approach to building effective legal prompts: Context, Role, Action, Format, Tone. Developed for Lawra to help legal professionals communicate effectively with AI systems.

AI tools designed for organizational use with enhanced security, data protection, access controls, and compliance features — as opposed to consumer-grade tools.

The degree to which an AI system's decision-making process can be understood by humans. Critical in legal applications where decisions must be justified and appealable.

The process of further training a general-purpose AI model on specialized data (e.g., legal texts) to improve performance in a specific domain.

AI systems capable of creating new content — text, images, code, audio — rather than simply analyzing existing data. ChatGPT, Claude, DALL-E, and Midjourney are examples.

When an AI model generates information that sounds plausible but is factually incorrect or entirely fabricated. In legal contexts, this commonly manifests as fictitious case citations or inaccurate statutory references.

A type of AI trained on vast amounts of text data that can generate, summarize, translate, and analyze human language. Examples include GPT-4, Claude, and Gemini.

A subset of AI where systems learn patterns from data rather than being explicitly programmed. The foundation for most modern AI tools.

A subfield of AI focused on enabling computers to understand, interpret, and generate human language.

The text input provided to an AI system to guide its response. In legal practice, prompt quality directly affects output quality.

The skill of crafting effective prompts to achieve desired AI outputs. Involves specifying context, role, instructions, format, and constraints.

A technique that enhances AI responses by retrieving relevant information from a curated knowledge base before generating output, reducing hallucination and improving accuracy.

The use of AI and machine learning to assist in reviewing large volumes of documents during litigation discovery. Court-approved in many jurisdictions.

The basic unit of text that an LLM processes. Roughly equivalent to 3/4 of a word in English. Models have token limits that constrain input and output length.

The dataset used to train an AI model. For LLMs, this typically includes books, websites, academic papers, and other text. The composition of training data affects model behavior and biases.

The neural network design underlying modern LLMs. Introduced in 2017, transformers process text by attending to relationships between all words in an input simultaneously.