Understanding the Generative AI Value Chain: A Harvard Business School Case Study

Generative AI Value Chain

This document outlines the core components and processes involved in the Generative AI value chain, as presented by Harvard Business School.

Understanding Generative AI

Generative AI refers to a type of artificial intelligence capable of creating new content, such as text, images, or audio, in response to user prompts. Prominent examples include ChatGPT and Bard for text generation, and DALL-E and Midjourney for image generation.

The Generative AI Process

1. Training:
   • Generative AI models learn by absorbing vast amounts of relevant data. This can include public domain books or extensive internet text.
   • The goal is to learn the underlying structure of the desired output.
2. Model:
   • At the heart of any generative AI system is the model, a mathematical representation of learned patterns.
   • The model's architecture, which is the theoretical organization of parameters in an artificial neural network, determines its structure.
3. Training Data:
   • A massive collection of examples relevant to the AI's task is crucial for learning.
4. Pre-training:
   • During this phase, the model adjusts its parameter-weights based on its architecture.
   • Prediction quality improves through numerous iterations over large datasets.
5. Fine-tuning:
   • The pre-trained model is further refined through a fine-tuning process.
6. Hardware Requirements:
   • Training AI systems demands specialized hardware, such as GPUs in data centers.
   • These systems consume significant amounts of electricity due to heavy, massively-parallel computational loads.
7. User-Facing Guardrails:
   • Commercial AI companies implement additional user-facing guardrails to prevent the generation of undesirable content.
8. Inference:
   • The trained model can then be used for inference by developers via an API or by users through an application.

Exhibit 1 (not provided in the text) is mentioned as an overview of the generative AI value chain.

Product Details

• Product #: 724355
• Pages: 18
• Publication Date: July 17, 2023
• Source: Harvard Business School
• Price: $8.95 (USD)

Available Formats and Languages

The product is available in various formats, including PDF, Audio MP3, Audio M4A, Audio CDROM, Audio Cassette, Bundle, DVD, Event Live Conference, Event Virtual Conference, Word Document, Electronic Book, Enhanced Electronic Book, ePub, Financial, Ebook, Hardcover/Hardcopy, Hardcover/Hardcopy (Color), Hardcover/Hardcopy (B&W), Web Based HTML, Kit, License, Magazine, Mobi, Multimedia CDROM, Multimedia Windows Media, Paperback Book, Powerpoint, Paperback/Softbound, Paperback/Softbound (Color), Paperback/Softbound (B&W), Registration Fee, Short Run, Subscription, Service, Video CDROM, Video DVD, Video Flash, Video VHS (NTSC), Video VHS (PAL), Video Real Player, Microsoft Excel Spreadsheet, XML, and Zip File.

Languages available include English, Spanish, Chinese, Danish, French, German, Japanese, Portuguese, Polish, Russian, Slovak, and Traditional Chinese.

Copyright Permissions

Copyrighted PDFs are for individual use only. To obtain copyright permission for team sharing, one copy per user must be purchased. Quantity pricing is available for bulk purchases.

Related Products

The document also lists related products:

• Bring Human Values to AI
• Generative AI-nxiety
• HubSpot and Motion AI (B): Generative AI Opportunities

Conclusion

The "Generative AI Value Chain" case study provides a comprehensive overview of the technology, its development process, and its commercial implications, highlighting the journey from data training to user inference.