Conversational AI Plugin Manager for Policy-Based Message Moderation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conversational AI tools like ChatGPT may respond inappropriately due to biased training data, leading to reputational damage and reliability concerns for organizations, and existing AI governance models are not designed to manage inappropriate behavior effectively.
Innovation Solution
A system and method for content management using a plugin manager to intercept and modify conversational AI messages based on organizational policies, enabling real-time moderation and ensuring compliance with AI governance frameworks, allowing multiple plugins to be invoked in parallel or sequentially.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conversational AI tools are trained using large language models, then they can provide beneficial features and functionality, but they may respond rudely or inappropriately due to biased training data
Solution Approach 1:
The patent introduces an intermediary system that sits between the user and the conversational AI tool. This intermediary intercepts messages, determines appropriate plugins based on policies, processes messages through selected plugins, and outputs modified messages. This mediator layer ensures that the AI tool's responses are filtered and moderated according to organizational policies, resolving the contradiction between functionality and response appropriateness.
Solution Approach 2:
The patent implements preliminary action by pre-defining policies and selecting plugins before messages are processed. The system determines one or more plugins to receive intercepted messages based on stored policies, and selectively forwards messages to these pre-selected plugins for processing. This preliminary setup ensures that appropriate moderation actions are taken before inappropriate responses can be generated, maintaining both functionality and reliability.
2Ease of operation
If conversational AI tools are used without content management, then they can operate freely, but they may cause reputational damage and reliability concerns for organizations
Solution Approach 1:
The intermediary system acts as a protective layer that allows the conversational AI tool to operate freely while filtering out harmful outputs. The system intercepts messages, processes them through policy-based plugin selection, and outputs modified messages that prevent reputational damage. This mediator approach maintains operational freedom while eliminating harmful factors.
3Reliability
If a plugin manager is introduced to intercept and modify messages, then content management and compliance are improved, but system complexity increases
Solution Approach 1:
The patent implements a universal plugin manager that handles multiple functions: intercepting messages, determining appropriate plugins based on policies, selectively forwarding messages to plugins, and outputting modified messages. This multi-functional component consolidates what could be multiple separate systems into a single versatile manager, improving content management while minimizing the increase in system complexity.
4Reliability
If multiple plugins are invoked in parallel or sequentially, then comprehensive policy enforcement is achieved, but processing time increases
Solution Approach 1:
The patent implements dynamic plugin invocation by selectively forwarding messages to one or more plugins based on policy determination. The system can invoke plugins in parallel or sequentially depending on the specific message and policy requirements. This dynamic approach ensures comprehensive policy enforcement while optimizing processing time by avoiding unnecessary sequential delays when parallel processing is appropriate.
Data Source
AI summary
Computing platforms, methods, and storage media for content management for a conversational artificial intelligence tool are disclosed. Exemplary implementations may: intercept, by an apparatus and from a data communication channel, a conversational AI message associated with the conversational AI tool; determine, by a plugin manager at the apparatus and based on stored policies, one or more plugins to receive the intercepted conversational AI message; selectively forward, at the apparatus, the conversational AI message to the one or more plugins for processing; and output, by the plugin manager at the apparatus, a modified conversational AI message based on the processing at the one or more plugins. Exemplary implementations may be configured to assist with one or more of: automatically moderating content sent to and from conversational AI solutions, assisting with ensuring responsible content and AI interactions, flagging undesirable intentions, and automating model governance.