Technological Readiness Versus Disruption: A Framework for Assessing Distinct Artificial Intelligence Policy Strategies

Abstract

Artificial intelligence (“AI”) is here and rapidly advancing. As the 2024 Thrower Symposium of the Emory Law Journal demonstrated, developments in the integration of AI to all aspects of life raise fundamental challenges for legal scholars and policymakers about the desirability and legitimacy of algorithmic decision-making, the societal impact of automation, and the liability, ownership, and regulatory puzzles created by substituting human actions with AI systems. This Article provides a framework for analyzing the readiness and desirability of contemporary technological innovation while continuing to address the effects and impact of technological shifts on the labor market, social welfare, and democracy. Drawing from examples of AI deployment in healthcare, transportation, public administration, and employment, this Article shows that technological readiness, including its comparative effectiveness, cost, and safety, can and should be measured and assessed separately from reforms targeting the economic and political implications of AI deployment. 

The Article delineates that, first, it is critical to adopt a comparative lens on AI versus human performance, asking whether the adoption of AI systems will nonetheless mitigate those risks and harms compared to current systems. Second, it is critical to distinguish between situations when AI is creating a new problem and when AI is simply reflecting existing societal problems back to us, when bias is detected in algorithmic recommendations. These comparative assessments are complex and nuanced; they must be answered across several axes, including consistency, efficiency, accuracy, scalability, accessibility, ability to detect and correct failure, traceability, explainability, and accountability. Because of this complexity, this Article suggests that a decontextualized attempt to regulate AI across sectors and applications is likely misguided. Moreover, such complexity requires a rich regulatory toolbox— including both command-and-control rules and public-private collaborative initiatives—drawing on the extensive literature and experience regarding effective governance. Third, separate from the assessment of the readiness of the technology for its intended purpose, policymakers must proactively evaluate and address the societal effects of the deployment of AI, including labor market disruptions, wealth distribution, consumer choices, and market competition. Thus, the Article introduces a two-step analysis of AI’s readiness and societal effects with research and data that is sector specific, establishing nuanced rational ongoing dialogue and policy. 

The two-step analysis would begin with an assessment of an AI application from the perspective of accuracy, consistency, speed, safety, costs, non-bias, system failure detectability and correction, scalability and access, comparative to a human decision-maker or a nonautomated system. The second step moves to assessing the impact of the adoption of the new application on the labor market, market competition and concentration, and other socioeconomic and environmental effects.

Keywords: artificial intelligence, AI, regulation, automation, decision-making, technical performance, societal impact, disruptive technology, bias, ai for good