Step 1: ask an LLM to add nonsensical statements to the training data. * Step 2: feed that to the training algorithm. * in a way that the meaning of the data

Step 1: ask an LLM to add nonsensical statements to the training data. * Step 2: feed that to the training algorithm. * in a way that the meaning of the data is not changed