AWS announces the launch of source document references for Automated Reasoning policies, simplifying the task of reviewing and refining an Automated Reasoning policy. Automated Reasoning checks uses formal verification techniques to validate that content generated by foundation models is compliant with an Automated Reasoning policy. Automated Reasoning checks deliver up to 99% accuracy at detecting correct responses from LLMs, giving you provable assurance in detecting AI hallucinations while also assisting with ambiguity detection in model responses.
To create Automated Reasoning policies, users upload documents that describe the rules in a knowledge domain like HR policies or financial transaction approval guidelines. These documents are translated into a collection of formal logic rules and variables called an Automated Reasoning policy. With source document references, users can now review the generated policy rules and variables using references to content they are familiar with from the original document,
Test generation for Automated Reasoning checks is now available in the US (N. Virginia), US (Ohio), US (Oregon), Europe (Frankfurt), Europe (Ireland), and Europe (Paris) Regions. Customers can access the service through the Amazon Bedrock console, as well as the Amazon Bedrock Python SDK.
To learn more about Automated Reasoning checks and how you can integrate it into your generative AI workflows, please read the Amazon Bedrock documentation, review the tutorials on the AWS AI blog, and visit the Bedrock Guardrails webpage.
Categories:
Source: Amazon Web Services
Latest Posts
- Microsoft 365 Copilot: New AI Reader role for Agent 365 (read-only access) [MC1296473]
- Microsoft Teams: Retirement of Together mode [MC1296478]
- Microsoft Purview | Data Security Investigations: Introducing new custom examination focus areas [MC1296479]
- Microsoft 365 Copilot (Premium): Teams meetings as a reference in Copilot Notebooks [MC1296488]
