A year and seven months after its creation, the Framework for Artificial Intelligence, which regulates the AI sector in Brazil and establishes principles for the use of this type of technology by companies, is ready to be voted on in the Chamber of Deputies. The text also provides that the technological field cannot be regulated by the federal government, for the time being. But it is up to the Union to propose rules to the AI sector.
The initial project of the Marco da Inteligência Artificial is by deputy Eduardo Bismarck (PDT-CE). However, the text underwent changes until it reached the plenary of the Chamber of Deputies on 7 July. The person responsible for writing a replacement version of the original bill was the rapporteur of the texts by the Chamber’s Science and Technology, Communication and Informatics Commission (CCTCI), deputy Luisa Canziani (PTB-PR).
The bill defines AI as follows:
“A system based on computational process that, from a set of goals defined by humans, can, through data and information processing, performed independently of human action, learn to perceive, interpret and interact with the external environment , making predictions, recommendations or ratings”
The definition also covers concepts that can be developed with Artificial Intelligence, but that are not exclusive to the technological field, such as machine learning.
Government supports vote on text in House
The text modified by the rapporteur has support from the MCTI (Ministry of Science, Technology and Innovation), as indicated by the director of Science, Technology and Digital Innovation at the folder, José Gontijo. “The substitute for Deputy Luisa Canziani (PTB/PR) was a very good text, serene, principled. It gives some guidelines, gives the government scope to regulate some things, but it does not immediately generate an impact, preventing innovation,” he said at an event held by the portal Digital Convergence, promoted by Network Eventos.
The rapporteur for the Framework for Artificial Intelligence described three main purposes of the bill. The first would be the safeguard that it is not possible to regulate an incipient sector through “static” rules. Second, PL must not “reinvent the wheel”. Finally, legislation must seek to improve AI in favor of the human being. Canziani explains that technology must be centered on the human being and conform to fundamental rights.
AI landmark imposes limits on the performance of the Public Power
To regulate the performance of the Public Power, the deputy included provisions for subsidiary intervention. This excerpt of the text, included in article 6 of the PL, says that the Union should not create norms on Artificial Intelligence, except in exceptional cases, when the government’s opinion is absolutely necessary.
When the intervention of the Public Power takes place, it must take place through the competent bodies, considering the “regulatory framework of each sector”. Any action taken by the government must consider the risks of intervening in each type of AI system and the respective market.
The PL also points out that any conduct of the State on AI sectors and systems must undergo public consultation before being implemented. This consultation should take place mainly through the internet and with ample prior disclosure to ensure broad participation by society.
Finally, one of the paragraphs of the bill included by Deputy Luisa Canziani (PTB-PR) emphasizes that the Union should promote and encourage the development of artificial intelligence when these stimuli are of low risk, that is, they should not cause a significant impact within of the technological field.
With information: Digital Convergence