Google’s AI Governance white paper calls for regulatory clean up.

(I got lotsa white paper. And why is it always called a white paper?)

Google’s white paper on AI governance has been released (January 22) and is an imporant read given the source. It calls for “concrete, context-specific” government action on AI regulation with five areas for consideration:

-explainability standards
-fairness appraisal
-safety considerations
-human-AI collaboration, and
-liability frameworks

The paper explicitly seeks to move the discussion from theoretical and high level to a greater degree of specificity and granularity on the issues. Included in initial considerations are that existing legal codes and principles in many cases have sufficient capability of dealing with AI issues or at least provide a starting point for further governance. Additional oversight and guidance are still necessary, the paper posits, and it is suggestive of a need for global standards to avoid a race to the bottom and regionalism. Considerations of prior major technological or innovations that could be considered harmful if misused and governance responses are highlighted as case studies that may promote thinking on regulatory responses to the introduction and use of AI.

Concluding remarks include the aim of flexibility, respect for cultural differences, and focussing on self and co-regulatory responses. Pure reliance on corporations to set standards is noted as undemocratic.

I’ll provide a deeper anaylsis of some of the issues in the paper soon.