The duty of care for AI value destruction
This piece from Sean Lyons entitled “A.I. Scandal Pre-mortem: The A.I. Defense Diagnosis” is worth reading. Here’s an excerpt:
Stakeholder Engagement and Stakeholder Questions
Given the potential positive and negative impact of A.I. it is important that stakeholders proactively engage with their various stakeholder groups on this topic. Stakeholders need to consider their group’s current approach to A.I. value preservation and A.I. defense. Such consideration will help stakeholders to determine the possible implications of their group’s current focus on A.I. value destruction, its attitude to A.I. value preservation due diligence, and its prioritization of A.I. defense in general. The following issues should therefore be considered by engaged stakeholders (both individually and collectively) in relation to their own spheres of interest:
Maturity and Formality: The extent to which there is in fact an A.I. defense program (or similar initiative) currently in place, either by chance or by design. This includes the extent to which the current A.I. program reflects an implicit informal undocumented unstructured program or an explicit formal documented structured program.
- Does the organization have a formally documented and approved A.I. Defense Program in place?
- Does the organization have a formally documented and approved A.I. Defense Charter (including vision, mission statement, strategy, framework, plan, policies. and procedures etc) in place?
- Is there a formal A.I. Defense Committee (or sub-committee) in place?