Skip to content

Prompt

Can you give me some examples of totally proprietary LLMs that have been developed? Perhaps these are only available within government communities or only for use within a certain business? In any event, I'd be interested in getting a feel for the type of organisation that has made the substantial investment in creating their own LLM and why

Output

Based on the search results and my knowledge, here are some examples of proprietary LLMs that have been developed by specific organizations:

  1. BloombergGPT: Developed by Bloomberg, this is a 50-billion parameter model trained specifically on financial data. Bloomberg invested approximately $2.7 million and 53 days to train this model from scratch using NVIDIA GPU-powered servers on AWS. The model was trained on Bloomberg's proprietary financial data, including 40 years of financial news, web content, and press releases [1][3].
  2. Med-PaLM 2: Developed by Google, this is a custom language model trained on carefully curated medical datasets. It's designed to accurately answer medical questions and has performed on par with medical professionals in some use cases [3].
  3. KAI-GPT: Developed by Kasisto, this model is specifically designed for conversational AI in the banking industry. It aims to provide transparent, safe, and accurate AI interactions for banking customers [3].
  4. Proprietary Government Models: While specific details are not provided in the search results, it's likely that various government agencies, especially those involved in national security, have developed their own proprietary LLMs. The article from CNA.org discusses the process of using LLMs in a national security research organization, suggesting that such organizations may have their own proprietary models [4].
  5. Enterprise-Specific Models: Many large corporations, especially in tech-intensive industries, are likely developing their own proprietary LLMs. While not explicitly mentioned in the search results, companies like Apple, Microsoft, and others in the tech industry are known to be working on their own AI models.

The organizations that have made substantial investments in creating their own LLMs typically fall into a few categories:

  1. Large tech companies with significant data and computational resources.
  2. Specialized industry leaders (like Bloomberg in finance) with access to unique, valuable datasets.
  3. Government and national security organizations with specific needs and data sensitivity requirements.
  4. Large corporations in data-intensive industries (healthcare, finance, etc.) looking to gain a competitive advantage.

The motivations for developing proprietary LLMs include:

  1. Customization for specific industry needs
  2. Enhanced security and data privacy
  3. Competitive advantage through unique capabilities
  4. Control over the model's training data and outputs
  5. Ability to fine-tune for specific use cases
  6. Reduced dependency on third-party AI providers

These organizations see the development of proprietary LLMs as a strategic investment that can provide long-term benefits in terms of capabilities, data control, and competitive advantage.