Why Autonomous Vehicles Should Not Be Labeled ‘High-Risk AI’
Even though fully driverless robotaxis are not yet commonplace, supervised autonomous vehicles have become a familiar sight on roads, particularly in South Korea. Tesla, a U.S.-based company, dominates the global market for autonomous vehicles, generating profits while simultaneously gathering extensive driving data to enhance its AI systems. Notably, Tesla plans to offer its autonomous driving capabilities through a monthly subscription model, starting in mid-February.
Korea’s leadership in industrial robot density is largely due to its robust electronics and automotive sectors. The nation utilizes more robots in manufacturing than any other country, with automation deeply integrated into automobile production. The automotive industry is now transitioning to a model where AI not only builds vehicles but also operates them, while humans pay for the use of AI technologies. Tesla has redefined automobiles in this AI era as software delivered as a service.
Korea’s AI Basic Act
As of January 22, Korea’s AI Basic Act has established a global precedent for regulating high-risk AI operators. Although penalties for violations have been deferred, the law assigns responsibility to AI providers to manage risks, mirroring similar regulations proposed in the European Union (EU). However, the EU has recently postponed its own high-risk AI regulations, fearing that a regulation-centered approach may hinder its global competitiveness.
The EU’s initial draft aimed to manage risks associated with high-risk AI systems, focusing on protecting human life and fundamental rights. This plan, however, faced challenges with the emergence of general-purpose AI systems like ChatGPT, which complicate regulatory definitions and objectives.
The Regulatory Dilemma
The rapid evolution of AI technology, especially following the introduction of ChatGPT, has led to significant changes in how AI is embedded in daily life. However, Korea’s regulatory framework for high-risk AI remains closely aligned with the EU’s 2021 draft, subjecting the country to international scrutiny as the first to fully implement such measures.
Under Korea’s AI law, Level 4 autonomous vehicles that are being piloted, as well as robotaxis already commercialized abroad, would fall under high-risk regulations. While the intent to classify technologies like autonomous vehicles as high-risk is to protect human safety, this categorization risks undermining national competitiveness. This is especially concerning when humanoid robots are viewed as innovative while autonomous vehicles are relegated to high-risk classifications.
Competitive Landscape
Tesla continues to sell vehicles at competitive prices, amassing driver behavior data globally without additional costs, which accelerates its technological advancements. To remain competitive against such companies, Korea must reassess its definition of high-risk AI and adjust its regulatory approaches to better reflect domestic realities.
One potential alternative could be a system similar to that of Britain, where insurance companies cover victims in the event of an autonomous vehicle accident and later seek recourse from manufacturers. This could alleviate the immediate liability burdens on automakers. Moreover, policies that promote the development of high-risk AI applications could be more beneficial than early, restrictive regulations.
In conclusion, while safety and innovation must coexist in the realm of autonomous vehicles, labeling them as high-risk AI without careful consideration may hinder progress and competitiveness in the rapidly evolving technological landscape.
