Improvement of artificial intelligence systems for automated classification of legal precedent results

• Grant number: 23/06218-0
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: November 01, 2023
• End date: April 30, 2025
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Danilo Rocha Limoeiro
• Grantee: Danilo Rocha Limoeiro
• Company: Turivius Portais de Conteúdo Ltda (Filial)
• CNAE: Desenvolvimento e licenciamento de programas de computador não-customizáveis
• City: São Paulo
• Pesquisadores principais:( Últimos ): Breno Osvaldo Funicheli
• Pesquisadores principais:( Antigos ): Guilherme Kenzo dos Santos
• Associated researchers: Breno Osvaldo Funicheli ; Edgar Nogueira Valente de Lima
• Associated scholarship(s): 24/16486-4 - Labeling of general legal data for the enhancement of artificial intelligence systems for the automated classification of legal precedent outcomes., BP.TT

Abstract

Brazil has one of the most complex legal systems on the planet. To exemplify with some data, there are more than 120 million lawsuits in progress in the judicial courts (CNJ, 2022), more than 28 thousand current tax rules (TCU) and the equivalent of 75% of the GDP is in dispute in the national courts (Insper , 2022). As a result, the negative impacts on economic activity, via the so-called "Brazil Cost", are considerable. On the one hand, structural reform solutions do not advance at the necessary pace in the public sphere, but, on the other hand, new artificial intelligence technologies have enormous potential to provide greater accessibility and predictability of the Brazilian legal system. Thus, the alternative way to reduce legal uncertainty and the "Brazil Cost" is through the application of highly innovative technologies in this sector. A promising way to give accessibility and predictability to the Legal System is through the application of statistical methods to Law, which is called Jurimetry. Jurimetry classifies the results of judicial decisions (or legal precedents) and, with that, manages to give predictability to jurisprudential trends and quantify the uncertainties of future judicial decisions. Given this context, the objective of this project is to improve legal artificial intelligence systems, through the use of LLMs and transformers technologies, to increase the scope and accuracy of automated classifications of results of legal precedents in the tax and labor domains. As a result of the introduction of this innovation in the market, it is expected to increase Turivius' competitiveness based on the scalability of the company's current technologies, allowing to increase the scope of service to the problem identified in the market, which is currently solved on a low scale and at a low cost high unit. (AU)