Debates on the regulation of Artificial Intelligence are on the rise in Brazil

The Senate Committee of Jurists presented a proposal to replace the bill on Artificial Intelligence (AI) currently being processed in Congress. The new text, which, although considered an advance when compared to the previous one, is considered too rigorous and comprehensive, proposes recommendations that minimize impacts caused by AI systems.

It is called for this replacement to be the starting point of the process, opening space for further advances, with new rounds of debate, taking into account the effects it may have on the use of Artificial Intelligence in economic activities, particularly aiming at the implications for small and medium-sized companies.

For society to be directly involved in the decisions of the new proposal, it is necessary for the debates to include experts and productive sectors, with the aim of bringing a bias towards technological trends and the impacts of adoption for the benefit of society, investment attractiveness and the competitiveness of companies.

It is considered essential that there be a process of analysis and contributions from authorities in other fields of knowledge, especially computer scientists, mathematicians, researchers and other professionals who, in their daily lives, deal with different Artificial Intelligence models and skills. Favoring data mining is one of the positive recommendations of the substitute text, as it can have a greater impact on the market. 

The proposal presented addresses a governance perspective that establishes responsibility in case of non-compliance with the law, mandatory disclosure about the use of AI and the defense of human rights, ensuring that algorithms do not encourage forms of discrimination. In addition to the interesting approaches, there are opportunities for improvement, from the perspectives of experts.

Among the more than 900 pages of the report, the objective of protecting fundamental rights, ensuring the implementation of safe and reliable systems for the benefit of the human person, the democratic regime and scientific and technological development, stands out. However, when categorizing excessive risks, it brings up controversial points related to social scoring and the use of biometric identification in public spaces, these being the most sensitive points.

Therefore, one of the main criticisms of the report is the distance between the view of jurists in relation to the application of AI systems, considering that an association between risk, interpretability and explainability of models would make more sense. It is understood that it would make sense to limit the use of some artificial intelligence models in high-risk activities, or to have explainability as a requirement for general use of AI. 

Considering the constant and rapid technological advances, some scenarios proposed by experts in the science of AI state that this technology will be present in moments such as employment selection processes, that is, scenarios that offer greater risks, making it necessary to calibrate the weight of the law more intensely, measuring the algorithmic impact of a collective risk for society.

Some rights of individuals who are directly affected by Artificial Intelligence systems are:

– Right to prior information regarding interactions with AI systems;

– Right to an explanation about the decision, recommendation or prediction made by this system;

– Right to challenge AI decisions or predictions that produce legal effects or significantly impact the interests of the affected party;

– Right to human determination and participation in decisions, taking into account the context and state-of-the-art technological development;

– Right to non-discrimination and correction of direct, indirect, illegal or abusive discriminatory biases;

– Right to privacyacity and protection of personal data in accordance with the relevant legislation.

An approach based on accountability and complemented by sectoral codes of conduct can also be effective in ensuring these fundamental rights when associating requirements with high-risk systems, and it is also necessary to incorporate existing legislation, such as the General Data Protection Law and the Marco Civil da Internet. 

The Network Rights Coalition says that the biggest flaw of the replacement involves the use of biometric identification, especially facial recognition, which must be preceded by federal law on the subject. With the absence of a general ban (aimed at public safety), the new PL is on the opposite side of campaigns that call for the use of the tool to be banned, which involve motivations ranging from structural racism to the selectivity of the Brazilian penal system. 

Another criticism made by the Coalition is regarding the gradation of risks, as it is considered mild, stating that the new text of the project only uses terms that recommend that some services be avoided or used with caution, without counting on specific prohibitions, leaving room for the adoption of technologies that are dangerous to the physical integrity of individuals or the exercise of rights. This becomes even more serious when related to systems for policing, arrests in the act, lethal weapons and emotion recognition.

The Committee of Jurists argues that the proposal is balanced, without encouraging punishments and maintaining the proportionality of sanctions, seeking that AI developers and suppliers see the more comprehensive law as a stimulus to reflect on ethics and responsibility in systems. According to the Commission, solutions and obligations are gradual and depend on the degree of risk presented.

Since the commission of jurists does not have the power to participate in legislative processes, it is up to the president of the Senate to continue processing the replacement. However, the debate needs to be more in-depth and seek a multisectoral basis.

Source: The Shift