On December 15, 2020, the Ministry of Science, Technology, Knowledge and Innovation opened for public consultation the first Draft of the National Policy on Artificial Intelligence, which, although it has received the first criticisms regarding its scope - a limited title only to Artificial Intelligence (AI) -, there are some points to consider regarding the public function in the field of taxes, especially considering the comparative experience in the European Union (EU).
The draft recognizes that Chile adhered to the recommendations of the AI Council of the OECD, an international organization that sets out five principles: (1) inclusive growth, sustainable development and well-being; (2) human-centered and justice; (3) transparency and explicability; (4) robustness, security and protection; and (5) accountability. This time I will only refer to the third principle on "transparency and explicability", known in the international literature as "algorithmic transparency".
In this regard, the comparative experience of the EU derived from the statements made at the 36th International Conference of Data Protection and Privacy Authorities (2014), the Opinion of the European Data Protection Supervisor (2016) and the Resolution of the European Parliament of March 14, 2017 on Implications of Big Data on fundamental rights, have recommended that the right of people to know the steps that the public administration follows in the use of AI or other technologies be recognized, explaining in a clear and understandable format how the algorithm or the process of automating the collected data works. Thus, then, recognizing the principle of transparency and explicability for the administrations is essential to subject the decisions of the public administration to due control and the right to challenge.
If we move to the tax field and, particularly, in cases where a tax administration uses AI or other technological processes to determine a tax, taxpayers have the right to know, in a clear and understandable format, the algorithm or automation process of data used to calculate taxes, especially if the data collected by this tool is not reliable.
However, this brings with it another problem: what happens if the algorithm used by the tax administration is owned by an external company? Is the intellectual and industrial property of that individual affected? Undoubtedly, this generates a collision of fundamental rights, being a great challenge to establish solution mechanisms against the new National Policy on Artificial Intelligence proposed by the Government. The literature offers solutions, but, first, said policy must clearly recognize, among other principles, algorithmic transparency.