Costa Rican lawmakers have taken an unconventional approach to drafting legislation on regulating artificial intelligence (AI) by enlisting the help of ChatGPT, OpenAI’s chatbot. The members of Congress instructed ChatGPT to “think like a lawyer” and create a bill in accordance with the country’s constitution. While the move has generated both positive and skeptical reactions, it highlights the growing urgency among legislators to establish AI regulations.
The resulting bill, recommended by ChatGPT, proposes the establishment of an institution responsible for regulating AI systems. This institution would prioritize principles such as accountability, explainability, bias prevention, and the protection of human rights. The bill was introduced in May and is currently undergoing public discussions before it progresses to the parliamentary commission for further deliberation.
Congresswoman Vanessa Castro, who spearheaded the bill, acknowledged the mixed responses it has received. She emphasized that the experience has underscored the crucial role of human involvement in shaping AI legislation. Costa Rica joins a growing list of Latin American countries, including Mexico and Peru, that are actively engaging in discussions and enacting laws to regulate AI.
However, some lawmakers, like Johana Obando, expressed reservations about the bill drafted by ChatGPT. Obando argued that the text contained fabricated statistics and references to the Costa Rican constitution. Her primary concern, shared by others in the region, is that AI legislation tends to lack substance and enforceability, often being viewed as a mere “list of good wishes.”
Latin American legislators draw inspiration from the European Union’s AI Act, which incorporates strict rules prohibiting the use of AI in biometric surveillance and calls for transparent disclosure of AI-generated content. In Brazil, where AI regulation has been the subject of intense debate for years, multiple bills are pending in Congress. One proposed framework aims to ban AI systems that may cause harm or target marginalized populations.
The need to combat bias and discrimination in AI systems is a central theme across the region. Advocates stress the importance of preventing facial recognition systems from leading to the disproportionate targeting and arrest of marginalized communities. Additionally, they highlight the necessity of addressing discrimination in automated hiring systems that disadvantage racial minorities. Calls for the right to review algorithmic decisions and potential reparations for individuals harmed by AI systems are also gaining traction.
A common thread in the discussions on AI regulation in Latin America is the desire to foster local innovation and competition with multinational corporations. Brazil, for instance, is considering a regulatory sandbox that would allow local businesses to experiment with AI technology under government authorization. Experts argue that promoting the development of homegrown AI products is crucial to combat the dominance of a few American multinational corporations.
Furthermore, experts emphasize the need for AI systems in the region to be culturally and linguistically tailored to Latin American contexts. They advocate for meaningful participation of Latin Americans in research and development processes, beyond serving as mere suppliers of data.
The engagement of Costa Rican lawmakers with ChatGPT to shape AI regulation marks a significant step in the region’s efforts to navigate the complexities of this rapidly evolving field. The ongoing discussions and legislative initiatives reflect a commitment to address the potential risks and challenges associated with AI while ensuring its ethical and responsible implementation.
