ChatGPT produces hallucination that Brazilian judge turned into legal decision
Today, journalist Lauro Jardim published the news about the case of the Brazilian judge who used ChatGPT to render a judicial decision. The judge is being sued by the CNJ (public agency that in Brazil regulates the activity of judges and judges the infractions they commit), because the decision contains hallucinations. Artificial intelligence invented a judicial precedent to justify the decision made and this is completely irregular.
Well, well, well... Whoever warns of dangers is a friend to people.
This incident, which will destroy the reputation of the judge involved and compromise that of the Brazilian Judiciary, would not have happened if the CNJ had already judged the administrative case that I myself filed asking that body to prohibit judges from using ChatGPT to make judicial decisions.
The technical report prepared in the process suggested that the CNJ authorize the use of that technological resource with a series of restrictions.
However, that process is awaiting resolution. The use of artificial intelligence of this type to automate decisions has many supporters in Brazil's justice system. Recently the president of the Supreme Court said that he will request the creation of a "legal ChatGPT" to be used by judges. In the State of São Paulo, the Court has already created a pilot project and made it available to judges.
And now the CNJ will have to judge this judge without there being any parameters for this to be done. Well done... If the CNJ had responded to a lawyer's request, it would have saved this judge's reputation and preserved that of the Brazilian Judiciary.