AI System Check Program Prioritizing Ethical Risk Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI system checklists and risk chain models do not provide clear guidance on how to implement ethical risk assessments, making it difficult for developers and providers to identify and manage ethical risks associated with AI systems, especially during system updates.
Innovation Solution
An AI system check program that acquires relational information about AI system components and stakeholders, prioritizes these based on data merging correlations, and outputs a checklist of relevant ethical check items to facilitate efficient risk assessment and management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If checklists from AI ethics principles and guidelines are applied to AI systems, then ethical risk recognition is improved, but the complexity of materializing these checklists and the burden of person-hours increases
Solution Approach 1:
The patent segments the comprehensive AI ethics checklist into multiple categories (data collection, data processing, model development, deployment, monitoring) with specific check items in each category. This segmentation allows developers to focus on relevant sections rather than materializing the entire checklist, reducing complexity while maintaining ethical risk recognition coverage.
Solution Approach 2:
The patent provides pre-materialized checklist templates based on established AI ethics principles and risk chain models. These templates are prepared in advance with structured formats and guidance, eliminating the need for developers to create checklists from scratch and significantly reducing the person-hours required for checklist materialization.
2Reliability
If risk chain models are used to organize risk constituent elements, then risk control examination is improved, but the burden of mapping risk elements to AI system components and stakeholders increases
Solution Approach 1:
The patent creates a universal mapping framework that can be applied across different AI system types and configurations. The framework provides standardized templates for mapping risk constituent elements to various AI system components and stakeholder roles, allowing the same methodology to serve multiple purposes and reducing the time required for custom mapping in each project.
Solution Approach 2:
The patent provides pre-defined mapping templates and examples that can be copied and adapted to specific AI systems. Instead of creating mappings from scratch, developers can replicate established mapping patterns from the templates, significantly reducing the time and effort required to map risk elements to their specific system components and stakeholders.
3Adaptability or versatility
If AI system configuration is updated, then system functionality is improved, but the efficiency of re-examining ethical risks decreases
Solution Approach 1:
The patent implements a feedback mechanism where the checklist evaluation results are fed back into the AI system development process. When system configuration is updated, the feedback from previous checklist evaluations helps identify which specific areas require re-examination, allowing for targeted rather than comprehensive re-assessment and maintaining efficiency during iterative development.
Solution Approach 2:
The patent makes the checklist evaluation process dynamic and adaptive to system changes. The framework allows developers to selectively re-evaluate only the portions of the checklist relevant to the specific configuration updates made, rather than requiring complete re-examination of all ethical risk areas, thus maintaining productivity during system evolution.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
An artificial intelligence(AI) system check program comprising instructions which when executed by a computer, cause the computer to execute processing including: acquiring a plurality of pieces of relational information that each includes at least two attributes among the attributes of a class of an object person, the attributes of the class of processing, and the attributes of the class of data, in which the relational information is determined based on a configuration of an AI system; designating priority of the plurality of pieces of relational information, based on whether or not each of the plurality of pieces of relational information is correlated with merging of the data; and outputting one or a plurality of check items selected based on the designated priority from among a plurality of check items each associated with the attributes, as a checklist for the AI system.