Follow Work Different With AI!
Work Different With AI
a metaphorical representation of a knowledge labyrinth, with concentric circular layers filled with people engaging in various academic and business activities. In the center, a man conjures a lightbulb, a universal symbol of an idea or insight. Surrounding him are individuals who are studying, discussing, teaching, and presenting, all against a backdrop of towering pillars and walls inscribed with alphanumeric codes, suggesting a complex repository of information or a library. The setting is bathed in a warm glow, highlighting a bustling atmosphere of intellectual pursuit within a vast expanse of knowledge.

How to Train Generative AI Using Your Company’s Data

By

WorkDifferentWithAI.com Academic Paper Alert!

Written by Tom Davenport, Maryam Alavi

Category: AI Strategy & Governance

Article Section: AI Strategy and Governance; Best Practices for AI Adoption

Publication Date: 2023-07-06T12:05:29Z

SEO Description: Discover how to enhance AI training with proprietary company data for competitive innovation and knowledge application.

Davenport, Tom, and Maryam Alavi. “How to Train Generative AI Using Your Company’s Data.” Harvard Business Review, 6 July 2023, https://hbr.org/2023/07/how-to-train-generative-ai-using-your-companys-data.

AI-Generated Paper Summary

GPT4-Turbo API

The article, authored by Tom Davenport and Maryam Alavi, discusses the importance of leveraging a company’s proprietary knowledge to maintain competitiveness and foster innovation in a volatile business environment. It addresses the challenge many organizations face in effectively organizing and deploying their comprehensive knowledge, which is often dispersed across various sources and mediums within the company. The authors explore the potential of companies using ChatGPT and other large language or image models, noting that while these AI systems demonstrate a remarkable ability to articulate complex ideas, they primarily rely on internet-based information and lack the capability to handle prompts or questions involving proprietary content. The article suggests that companies are experimenting with these technologies to potentially embed their intellectual capital into large language models. The aim is to better manage and apply their knowledge assets, which is crucial for organizational innovation and staying ahead in today’s fast-paced digital landscape.

Claude.ai Full Article Analysis

Here is my analysis of the HBR web post “How to Train Generative AI Using Your Company’s Data”:

Novelty:

  • The post provides a timely overview of different techniques companies can use to train large language models on their proprietary data, including training from scratch, fine-tuning, and prompt-tuning. This is a novel application of generative AI that is just emerging.
  • It highlights innovative examples of companies like Bloomberg, Google, Morgan Stanley, and Morningstar that are early adopters of training generative models on their own data. These are novel use cases.
  • The discussion of legal, governance, and user behavior issues related to training generative AI on company data provides a novel perspective on potential risks and how to mitigate them.

Commercial Applications:

  • Training generative AI on company data could enable new customer service applications like the Fortune 500 software company example, improving productivity and customer satisfaction.
  • Applications for financial services firms like personalized investment advice based on proprietary research at Morgan Stanley and low-cost Q&A like the Morningstar example.
  • Capturing employee knowledge before departure and distributing it across the company, as well as training new hires.
  • Automating information search and retrieval for domains like legal case research.
  • New product development applications based on recombination of existing proprietary knowledge.

In summary, this provides a timely overview of an emerging application of generative AI that has many promising commercial use cases and examples. The discussion of implementation challenges and risks provides a balanced perspective.

Keywords

Generative AI, proprietary knowledge, organizational innovation, knowledge assets, large language models

Author’s Abstract

Leveraging a company’s proprietary knowledge is critical to its ability to compete and innovate, especially in today’s volatile environment. Organizational innovation is fueled through effective and agile creation, management, application, recombination, and deployment of knowledge assets and know-how. However, knowledge within organizations is typically generated and captured across various sources and forms, including individual minds, processes, policies, reports, operational transactions, discussion boards, and online chats and meetings. As such, a company’s comprehensive knowledge is often unaccounted for and difficult to organize and deploy where needed in an effective or efficient way.

Read the full paper here

Last updated on November 19th, 2023.