Resumo

Error is a key element in machine learning, as every modern widely used machine learning algorithm performs some task to minimize – and not to nullify – a loss function, which basically measures the difference between the output of the predictive model and the function it tries to approximate. In that sense, machine learning algorithms, as they are today, will always make mistakes, even with large well-constructed datasets. This limitation, by itself, poses a challenge for lawmakers and jurists, as it may influence regulatory efforts or the possible usage of automated systems in reallife applications that may impact the legal field.