 Название: AI Frameworks Enabled by Blockchain: Creating Trustworthy and Responsible AI Using Distributed Ledger Technology
Автор: Vikram Dhillon, David Metcalf, Max Hooper
Издательство: Apress
Год: 2025
Страниц: 564
Язык: английский
Формат: pdf
Размер: 11.2 MB
Blockchain technology offers a powerful foundation for building trust, privacy and verifiability into AI frameworks. This book will focus on how a blockchain can enable AI frameworks and applications to scale in a responsible fashion, reshaping the future of numerous industries from financial markets to healthcare and education.
You’ll see that in the next wave of AI products, blockchain can provide a “Trust Layer,” a fundamental feature previously only implemented for parties within a blockchain network. The provable consensus algorithms and oracles previously implemented in blockchains can be extended to autonomous agents that are integrated with Large Language Models (LLMs) and future applications.
Finally, you’ll learn that safety is a major concern for practical applications of AI and blockchain can help mitigate threats due to the decentralized nature. As such, there will be significant discourse on how blockchain can provide enhanced security against prompt injections, LLM-hijacking for dangerous information and privacy. These ideas were studied rigorously when large financial institutions were releasing their own blockchains and distributed ledger protocols with a heavy focus privacy.
AI is undergoing a Cambrian explosion this year with foundational models emerging for all major domains of study, however, most such models lack the capacity to externally validate for the “correctness” of a fact, or reply made by the LLM. Similarly, there are no definitive methods to distinguish between meaningful insights and hallucination. These challenges remain at the forefront of AI research, and AI Frameworks Enabled by Blockchain aims to translate technical literature into actionable and practical tips for the AI domain.
Instruction fine-tuning further refines a LLM to generate higher-quality responses and address a user’s request when prompted with specific instructions. For instance, we can take a LLM that is intended to answer questions and summarize tests and fine-tune its behavior so that it becomes adept at coding in Python and generating coherent, targeted responses to coding questions. Llama, the LLM from Meta, has undergone such fine-tuning to produce coding models for Python that generate reliable, relevant, and targeted outputs.
What You Will Learn:
· Bring a layer of accuracy to generative AI where a non-generative component behaves as guardrails
· Protect users from harmful biases as well as hallucinations.
· See how blockchain plays a role in aligning AI with human interests.
· Review use-cases and real-world applications from parties that have invested a significant amount in building technology stacks utilizing both.
Who This Book Is For:
Enterprise users and policy makers in the field of Professional and Applied Computing.
Скачать AI Frameworks Enabled by Blockchain: Creating Trustworthy and Responsible AI Using Distributed Ledger Technology
|
Навигация
Вход на сайт
Реклама
AI Frameworks Enabled by Blockchain: Creating Trustworthy and Responsible AI Using Distributed Ledger Technology 
