Large Language Model
Definition
Massive AI models trained on vast amounts of text data that can understand and generate human language, enabling advanced natural language processing.
Use Cases
- Duolingo: AI-powered conversation practice and personalized language learning features (Duolingo Max). — Integrated GPT-4 via Azure OpenAI Service into product features such as roleplay conversations and explanations, with application-layer guardrails and prompt design to keep responses aligned to learning goals. (Launched premium AI features that enhance interactivity and personalization, supporting product differentiation and new subscription value.)
- Khan Academy: AI tutoring assistant (Khanmigo) to help students learn through guided questions and explanations. — Built an LLM-powered tutoring experience using GPT-4 through Azure OpenAI Service, adding safety policies, teacher/parent controls, and instructional prompting to encourage step-by-step learning rather than just answers. (Delivered an AI tutor experience used for learning support and classroom pilots, expanding access to guided practice while emphasizing safety and pedagogy.)
- Morgan Stanley: Internal knowledge assistant for financial advisors to search and summarize firm research and documents. — Used GPT-4 with a retrieval-augmented generation (RAG) approach to ground responses in approved internal content, with access controls and auditing to meet compliance needs. (Improved speed of information retrieval for advisors and increased usability of internal research content through conversational search.)
Provider Equivalents
- AWS: Amazon Bedrock
- Azure: Azure OpenAI Service
- GCP: Vertex AI (Gemini models)
- OCI: OCI Generative AI
Frequently Asked Questions
- What's the difference between a Large Language Model (LLM) and generative AI?
- Generative AI is the broad category of systems that create new content (text, images, audio, code). An LLM is a specific type of generative AI focused on understanding and generating text (and often code). In practice, LLMs are one of the most common engines used to build generative AI chatbots and writing tools.
- When should I use a Large Language Model?
- Use an LLM when you need natural-language capabilities such as chatbots, document Q&A, summarization, drafting emails or reports, extracting structured data from text, or code assistance. LLMs work best when you can tolerate probabilistic outputs and you can add guardrails (e.g., grounding with your documents, validation rules, and human review) for high-stakes workflows.
- How much does a Large Language Model cost?
- Costs usually depend on usage: the number of input and output tokens (text length), the model chosen (larger models cost more), and features like fine-tuning, embeddings, or tool/function calling. Additional costs can come from retrieval (vector databases), data storage, networking, and monitoring. For predictable spend, many teams set rate limits, cap response length, cache frequent answers, and use smaller models for simpler tasks.
Category: ai-ml
Difficulty: advanced
Related Terms
See Also