Chapter 2. Blockchain Controls for AI
As you learned in Chapter 1, there are a wide variety of potential blockchain touchpoints for AI. However, nearly every process can be traced and verified by implementing one of the four controls explored in this chapter. You will learn to plan and implement each control by determining potential AI scenarios for each and explaining how it might be addressed using Hyperledger Fabric.
Four Blockchain Controls
When approaching your AI factsheet and thinking about how you might design a fact flow system around it, you may find that thinking of your architecture in terms of these four categories of controls simplifies your planning. Table 2-1 illustrates these four controls and shows how they are tied together by governance. The four blockchain controls for AI—pre-establishing identity and workflow, distributing tamper-evident verification, governing, instructing and inhibiting intelligent agents, and showing authenticity through user-viewable provenance—are interwoven. Governance, in the form of bylaws, smart contracts and consent, overlaps them all.
| Control | Impacts | Description |
|---|---|---|
| Control 1: pre-establishing identity and workflow criteria for people and systems | Participants, assets, and transactions | Includes criteria for telling humans apart from AI |
| Control 2: distributing tamper-evident verification | Databases, models, libraries, federated AI | Tests data against a cryptographic hash to detect ... |
Get Blockchain Tethered AI now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
