In March 2024, the European legislation on artificial intelligence (AI) was approved. Chile, which had previously pioneered constitutional regulation on neuro-rights, aims to catch up with a bill introduced in May 2024.

It is essential to recall some key concepts regarding AI and its scope: AI is a technology that, unlike other disruptive technologies in human history, effectively rivals human capabilities, even surpassing them in certain mental tasks. This potential to outperform humans in tasks considered exclusively human for thousands of years is what raises concerns and forms the basis for its regulation.

Broadly speaking, AI is a technology designed to simulate certain human mental processes. Since the human mind is seen as having control over human actions, AI stirs both fantasies and fears.

As a discipline, AI originated as an attempt to simulate certain human mental processes through artificial, technological means, specifically digital ones. Initially, it focused on tasks believed to be uniquely human, such as playing checkers and chess, making logical deductions, translating languages, solving puzzles, or finding the exit of a maze. AI also aimed to simulate medical experts, builders, and more. Later, it began automating well-defined human mental functions, like searching through archives or performing complex mathematical operations. All of this falls under the realm of traditional AI.

Today, what we refer to as AI is a digital technology that seeks to mimic biological neural networks, and it has succeeded in creating machines with powerful capabilities to simulate many human mental activities. What marks a qualitative leap is the development of machines that learn from experience—improving their performance by learning from both errors and successes. This is known in computing as “machine learning.” The achievements of this technology are already impressive, with systems that simulate and even surpass many human functionalities, such as pattern detection, searches, and predictions. In these tasks, the systems exhibit autonomy based on their core digital neural networks.

The regulatory bill to be discussed in the Chilean Parliament defines AI as “a set of systems based on machines that infer, from input information, certain output information, which may consist of predictions, content, recommendations, or decisions capable of influencing physical or virtual environments.” The law itself formally defines “AI systems,” but it does not clearly distinguish between AI and “AI systems.”

The purpose of the law is outlined in Article 1 as follows: “This law aims to promote the creation, development, innovation, and implementation of artificial intelligence (‘AI’) systems that serve humanity, are respectful of democratic principles, the rule of law, and the fundamental rights of individuals against the harmful effects that certain uses may cause.”

Notably, the text implicitly attributes human agency to these systems: “that are respectful of democratic principles…etc.” In reality, it is those who create and use these systems who should be respectful of democratic principles. It is crucial to emphasize this point, as it is not these new technical objects that should be required to respect certain principles, but the human agents involved in their deployment (no machine has agency on its own).

Article 2 explicitly excludes from this law’s regulations national defense, research activities, and open-source and free-license AI components. The rationale behind these restrictions is reasonable.

Article 5, which forms the core of the law, organizes a classification of systems based on the risk of harm they may cause (defined as “the combination of the likelihood of harm to natural persons, their health, safety, or fundamental rights, and the severity of such harm”). It establishes four risk categories: unacceptable risk, high risk, limited risk, and no evident risk, and proposes a set of requirements of varying intensity for each category.

The remainder of the law covers general regulatory matters related to innovation and digital issues rather than AI specifically, such as support for innovation, data management, intellectual property, and more.

In summary, it is highly significant that Chile is considering regulating AI, following the global trend on this issue. It is equally important to understand that, on the one hand, AI is part of a broader spectrum of digital transformations, and on the other hand, like all technology, AI is not just technological objects but also a know-how—a skill set that, as is well known, cannot be imported. This last point is particularly relevant for countries like Chile that are still in the process of shaping their identity in terms of socio-technical infrastructure and development.

It is the responsibility of everyone to outline how this new disruptive technology will be integrated into the social, educational, industrial, and service sectors of the country in a way that not only avoids harming people but also includes them in its developments. Finally, it is worth highlighting that the current bill has some gaps or lacks emphasis on critical issues that should be addressed, such as national digital sovereignty, the promotion of national research and the educational dimensions of the phenomenon, the protection of workers in the field, and ensuring equal access to these new technologies.