Brazilian authorities have confirmed that a federal judge in the northern state of ICA has been ordered to explain the reasons for the publication of a decision full of errors, co-authored by chatbot ChatGPT.
The National Council of Justice (CNJ) gave Judge Jefferson Rodriguez 15 days to explain a decision full of incorrect details about previous court cases and legal precedents, including the erroneous attribution of previous decisions to the Supreme Court of justice, the case records revealed.
Rodriguez admitted in the documents submitted to the supervisory authority that the decision was written jointly with a “trusted consultant” and artificial intelligence. He ignored the mistake, calling it “just a mistake” made by one of his subordinates, and blamed the “excessive workload facing the judges” for the mistakes.
The CNJ board claimed that the incident was the” first case of its kind ” in Brazil, which does not have laws prohibiting the use of artificial intelligence in judicial settings. In fact, the president of the Supreme Court plans to commission the creation of a” legal ChatGPT ” for use by judges, a project that is reportedly already underway in Sao Paulo.
Judges have been using AI-powered chatbots to guide their decisions for almost the entire period that they have been publicly available, despite their tendency to produce highly visible “hallucinations”, responses that have no basis in reality.
Colombian judge Juan Manuel Padilla Garcia, of the First Circuit Court of Cartagena, proudly credited ChatGPT with the decision he made in January on whether a child with autism should receive insurance coverage for medical treatment, qualifying the unusual research method while confirming that its responses were real. They were examined and”in no way intended to replace the judge’s decision”.
