Regulation and Cyborgs

<p><span>Artificial intelligence is an old technology made new by the popularity of generative AI applications. Although hybrid systems that combine humans with computer-based automation—which this Article terms “cyborgs”—have a long history, policymakers and legal scholars alike perceive AI as an isolated, monolithic, revolutionary technology that is devoid of context. That approach is a mistake. This Article’s core claim is that governance of cyborgs should situate AI in context, by the specific technology and use at issue, and draw lessons from the history of human-computation co-production for more effective and less burdensome oversight. AI is a foundational technology and should be regulated as such. The Article draws upon four extant examples of cyborgs for guidance: Boeing’s disastrous 737 MAX redesign, the Internal Revenue Service’s National Research Program Audits, consumer credit lending for home mortgages, and the Volkswagen engine that caused the Dieselgate emissions scandal.</span></p> <p>&nbsp;</p> <p>The history of cyborg systems teaches four vital governance lessons for AI. First, blended human-automation decisionmaking is increasingly ubiquitous, diverse, and inevitable in a world flooded with data. Second, normal accident theory shows that having a human in the loop can usefully create resilience, but only if the cyborg’s design enables that person to exercise meaningful autonomy in decisionmaking. Third, testing is the key to effective governance: hybrid systems are what they actually do. Fourth, humans are inextricably enmeshed in cyborgs, because only humans can exercise the moral judgment necessary to make values-laden choices about system architecture and policy. The Article applies these lessons to two AI current controversies, over copyright and defamation, to clarify how these doctrines have previously dealt with hybrid information production and what is at stake normatively in these debates for AI governance.</p>