| Abstract | The successful economic, social, and cultural integration of migrants is vital for sustainable global development. In Canada, the nonprofit settlement sector plays a crucial role in helping newcomers achieve this integration. However, as immigration targets have risen, this sector is experiencing growing operational pressures, which underscores the need for greater efficiency and innovative solutions, potentially through adopting reliable AI technologies. Increasingly, newcomers and service providers may adopt general-purpose generative AI tools such as ChatGPT to meet these challenges. Yet, these tools are not specifically designed for the settlement domain and could have harmful effects on immigrants and refugees. Our study investigates the risks associated with these tools to firstly caution against their indiscriminate use and secondly, to encourage more research and development in AI literacy programs and customized large language models that respect the preferences of affected communities. Importantly, such technologies must be designed to integrate seamlessly into the existing workflows of the settlement sector, ensuring human oversight, trustworthiness, and accountability. |
|---|
