LLM-Supported AI System

An LLM-Supported AI System is an GenAI-supported system that implements an LLM-based algorithm to solve an LLM-based task (which requires the use of a large language model).

• Context:
  • input: Large Language Model (LLM) Prompt.
  • It can (typically) be created by an LLM Engineering Team (that follows a LLM system development model).
  • It can (typically) be composed of LLM Pipelines (such as an LLM training pipeline).
  • It can (typically) make use of an LLM Technology (such as an LLM platform or LLM libraries).
  • It can range from being a Partially-Automated LLM-based System to being a Fully-Automated LLM-based System.
  • It can range from being a Passive LLM-based System to being a Active LLM-based System.
  • It can range from being a Batch LLM-based System to being an Online LLM-based System.
  • It can range from being a Personalized LLM-based System to being an Non-Personalized LLM-based System.
  • It can range from being a Zero-Shot LLM-Supported AI System to being a Few-Shot LLM-Supported AI System.
  • It can range from being an LLM-Supported Application to being an LLM-Supported Component.
  • It can range from being a LLM-based Demo-Ware System to being an LLM-based Application.
  • It can be implemented using an ML Framework (that can include an ML library, or ML tools).
  • …
• Example(s):
  • LLM-Supported Application, such as:
    • an LLM-supported Open-Domain Chatbot, such as: Github Copilot, OpenAI ChatGPT, Anthropic Claude, ...
    • an LLM-based Perspnal Assistant, such as: ...
    • an LLM-based Customer-Supporting System, such as: ...
    • an LLM-based Law-Supporting System, such as an LLM-based contract-focused analysis system.
    • an LLM-based Healthcare-Supporting System, such as an LLM-based Medical Diagnosis System or an LLM-based clinical trial system.
    • an LLM-based Education-Supporting System, such as: ...
    • ...
  • an LLM-based Information Extraction System, such as: ...
  • an LLM-based Text-Classification System, such as:
    • an LLM-based Sentiment Classification System.
    • ...
  • an LLM-based Content Generation System, such as:
    • an LLM-based Question Answering System, such as: ...
    • an LLM-based Document Summarization System, such as: ...
    • an LLM-based Term Definition Summarization System, such as: ...
    • an LLM-based Language Translation System, such as: ...
    • ...
  • an LLM-based Recommender System, such as: ...
  • an LLM-based Agent.
• Counter-Example(s):
  • a Predictive ML-based System.
  • a Rule-based AI System.
• See: ML-based System.

---

---

References

2023

• https://www.fiddler.ai/blog/four-ways-that-enterprises-deploy-llms
  • QUOTE: With the rapid pace of LLM innovations, enterprises are actively exploring use cases and deploying their first generative AI applications into production. As the deployment of LLMs or LLMOps began in earnest this year, enterprises have incorporated four types of LLM deployment methods, contingent on a mix of their own talent, tools and capital investment. Bear in mind these deployment approaches will keep evolving as new LLM optimizations and tooling are launched regularly.

    The goal of this post is to walk through these approaches and talk about the decisions behind these design choices.

  • 
    Four approaches to LLMs in production (AWS Generative AI Summit)
  • There are four different approaches that enterprises are taking to jumpstart their LLM journey. These four approaches range from easy and cheap to difficult and expensive to deploy, and enterprises should assess their AI maturity, model selection (open vs. closed), data available, use cases, and investment resources when choosing the approach that works for their company’s AI strategy. Let’s dive in.

2023

• chat
  • Q: An LLM-based production system refers to a system that utilizes large language models (LLMs) like OpenAI's GPT series to enhance, automate, or support various aspects of production or operations. These systems leverage the natural language understanding, generation, and reasoning capabilities of LLMs to perform tasks that typically require human intervention or expertise. While I'm not aware of any specific production systems called "LLM-based production systems," I can give you some examples of systems that utilize LLMs in their functioning:
    • Content generation: LLMs like GPT-4 can automatically generate content for blogs, social media, and marketing campaigns, reducing the time and effort required by human content creators.
    • Customer support: LLMs can be integrated into chatbots and support systems to handle customer queries, complaints, and concerns more effectively and efficiently.
    • Document summarization: LLMs can automatically summarize long documents, reports, or articles, making it easier for professionals to extract relevant information quickly.
    • Legal analysis: LLMs can be employed to review contracts, analyze legal documents, and provide insights into complex legal matters, saving time and resources for legal professionals.
    • Language translation: LLMs can be used to develop translation systems that accurately convert text from one language to another, making communication easier across different countries and cultures.
    • Sentiment analysis: LLMs can analyze customer reviews and feedback to identify positive or negative sentiments, helping businesses improve their products and services.
    • Medical diagnosis: LLMs can assist medical professionals by analyzing patient symptoms and medical history to suggest potential diagnoses and treatment plans.
  • These are just a few examples of how LLMs can be integrated into various production systems to enhance efficiency, reduce costs, and improve overall output. As AI research and development continues, LLMs will likely find even more applications across a wide range of industries.

---