Pie & AI:Suisse – Trustworthiness of AI models:Improving NLP with Causality