[TMP-010] A Transparent and Explainable Dialogue System for Immigration Services
This project developed an ethical chatbot integrating AI and legal expertise to guide asylum seekers while ensuring transparency and privacy.
Migration is a key concern for the European Union, requiring the integration of ethical, legal, and societal considerations. However, the complex legal rules around immigration and asylum make it challenging for migrants to assess their chances of obtaining protection. This micro-project explored how trustworthy AI can streamline asylum application processes. A multidisciplinary team of computer scientists and immigration law experts developed a chatbot to support and guide asylum seekers in Europe, without replacing judiciary experts or enabling “predictive justice.”
Building on the “Ethical Chatbots” project (Fazzinga et al., 2022), we enhanced our argumentative chatbot architecture to tackle this domain's complexity and leverage LLMs. The focus remained on data governance, privacy, transparency, explainability, and auditability.
Our work highlighted the importance of collaborating closely with domain experts. Besides understanding legal frameworks, the process highlighted the value of unwritten best practices known only to experienced practitioners. Additionally, access to asylum applications and court decisions is restricted for applicants' safety, limiting a data-driven approach. Current LLMs, trained on publicly available data, also lack the necessary information to provide meaningful answers, further emphasizing the need for expert-guided development.
We developed “ACME”, a prototype chatbot with the aim of supporting migrants in their requests for asylum. ACME is a hybrid architecture that combines a subsymbolic language understanding module based on NLP techniques and LLM, with a symbolic reasoning module based on computational argumentation.
The aim of the tool is to help migrants identify the highest level of protection they can apply for. This would contribute to a more sustainable migration by reducing the load on territorial commissions, Courts, and humanitarian organizations supporting asylum applicants.
Relevant properties ACME exhibits include: data governance and privacy thanks to its modular architecture; transparency and explainability thanks to argumentative reasoning; and the ability to integrate and reasoning with explicit, expert-made, formalized knowledge, ensuring auditability.
Tangible Outcomes
- Bettina Fazzinga, Elena Palmieri, Margherita Vestoso, Luca Bolognini, Andrea Galassi, Filippo Furfaro, Paolo Torroni (2024). “A Chatbot for Asylum-Seeking Migrants in Europe”. IEEE International Conference on Tools with Artificial Intelligence (ICTAI) http://arxiv.org/abs/2407.09197
- chatbot code https://github.com/lt-nlp-lab-unibo/ACME-A-Chatbot-for-Migrants-in-Europe
- Video demonstration of the tool: https://www.youtube.com/watch?v=P8iW7FOZTYM&feature=youtu.be
Partners:
- University of Bologna, Andrea Galassi, a.galasi@unibo.it
- University of Calabria, Bettina Fazzinga, bettina.fazzinga@unical.it
- University of Naples, Margherita Vestoso, margherita.vestoso@unina.it