Establishing Trust in AI: Blockchain​

Blockchain serves as an immutable record and enforcer for responsible AI development, ensuring models meet standards before deployment

Three key roles in auditable AI development:

• Assignee (data scientist) responsible for model development
• Tester who validates the work
• Verifier who provides final independent verification

Model development requires meeting predefined requirements around:

• Bias testing across protected classes
• Interpretability and explainability
• Ethical considerations
• Monitoring specifications
• Success criteria

No AI model can be released until all requirements are met and recorded on the blockchain, creating full transparency and accountability

Consumer trust in AI is low - surveys show 61% are wary of trusting AI systems and 73% perceive significant risks

Organizations need a single model governance standard to ensure consistent development practices across teams

Currently lacking unified standards - different requirements exist across jurisdictions and industries

Blockchain implementation allows for:

• Immutable record of model development decisions
• Clear documentation of requirements and success criteria
• Auditability of AI models
• Enforced governance processes

Focus should be on building interpretable, robust AI models rather than overly complex black boxes

Regulatory compliance will become increasingly important as AI oversight grows stricter globally