Into the next generation of digital protection: AI resiliency as a public responsibility

Even as the reliability of networks has risen, their control mechanisms up in the hierarchy of digital activities have become more vulnerable. Artificial intelligence algorithms are increasingly embedded in infrastructure and economic systems and their resiliency is essential for social and economic stability. This has led to widespread dystopic fears and defensive regulations, ignoring the considerable positives of AI-enhanced activities and institutions. AI resiliency problems include hardware failures, natural calamities, human error, software defects, and external attacks. AI networks of AI networks have emerged with high interdependence and complexity. Operations are often non-transparent ‘black boxes’ operating at lightning speeds, and hard to oversee or fix by humans. Most likely is a control of AI by other AI. This raises the question of human responsibility. The article examines various responses, including technology tools, managerial actions, self-regulation, and a role for government. The latter include rules evolving with technology and applications in a dynamic common law approach for liability, transparency, performance, market structure, interoperation, and more. Needed are principles for a ‘shared intelligence’ of humans with AI, with clear protocols for human overrides of AI. All this raises a new agenda for policymakers, managers, and researchers.