Skip to Main Content
You are now in the main content area

Generative AI, Migration, and Participation (GenAMP)

A collage that merges circuit board patterns with textile motifs in a grid-like background of alternating black, grey, and white. Two hand-drawn arms are on each side of the image, positioned as if gently pulling on thin, white strings that cross the image diagonally. The hands appear soft and somewhat translucent, contrasting with the rigid lines of the circuit board patterns behind them. The strings are woven through both the hands and the background, symbolising the connection between traditional weaving and modern technology. The overall colour palette features muted earth tones, including browns, beiges, and grays, creating a sense of both history and continuity between the natural and technological worlds..

Project Leads

Naimul Khan, Annie Wan

Team Members

Masoud Kianpour, Reza Samavi, Mireille Paquet

Theme: Citizenship and Participation
Research Cluster: Methods and Infrastructure Collaboration 
BD-branding-line-blue - 1

Objective

Generative AI, Migration, and Participation (GenAMP) is an interdisciplinary research cluster exploring the application of generative AI (genAI) in two domains:

1. GenAI applications in text, where Large Language Models (LLMs) are intended to be used to complete different immigration based tasks, e.g., (a) developing legal documents for immigrants with the help of GenAI, (b) developing chatbots to help provide settlement services to immigrants, and (c) developing tools for analysing policies that affect immigrant populations. 

2. GenAI applications on visual and immersive media (e.g., image synthesis, video synthesis, augmented/virtual reality), which simulate or represent various aspects of the migration experience. 

We will investigate how different GenAI systems (text, images, videos, AR/VR) portray migration, migrants and the immigration process as well as any inherent bias that might occur in these representations. How does GenAI shape public perceptions about immigration and how does it differ in this respect from content that has been generated by humans? We will also explore ways to design GenAI systems as participatory tools that aid newcomer communities, instead of creating further barriers with respect to digital literacy, language and cultural context. We will identify where the current technological limitations of AI safety mechanisms and content guardrails are when applied to migration sensitive contexts and how to create mechanisms that extend beyond simple surface level filtering to provide more profound, robust forms of safety. We will investigate the propagation of algorithmic bias present in generative models, through immigration systems, and what auditing frameworks are required in order to identify and eliminate these biases in each of the modes (text, image, video) within immigration-related applications.

BD-branding-line-blue - 1

Research Questions

The project will address technical issues with implications for the broader community through a consideration of three questions:

  1. How do algorithmic bias inherent in generative models affect how we frame the migration experience?
  2. Do genAI models produce offensive content (in a migration context), despite existing safeguards?
  3. How can we use the insights gained from investigating the internal decision-making processes used by GenAI systems to help create trustworthy GenAI tools for vulnerable populations?
BD-branding-line-blue - 1

Methodology

The research will blend together technical and qualitative social science approaches to investigate how GenAI interacts with issues related to immigration. On a technical level, the researchers will perform an analysis of GenAI outputs across modalities (text, images, videos), focusing on how AI models respond to immigration-related queries in different languages, cultural contexts and how those same models generate visual representations of people and situations related to migration. In addition to assessing how detectable bias is present in the output of GenAI produced by these models, the researchers will also be conducting audits to better understand what processes take place within these systems in terms of bias and how biased outputs are generated by AI, as opposed to seeing the system as a black box. 

The technical team will then partner with migration scholars and will engage the newcomers and settlement service provider communities in Canada through participatory design methods. By including these communities as co-designers, rather than as passive subjects, the evaluations of GenAI tools can be based on real-life experiences by the end-users of these tools. This aspect of the project will also incorporate a visual and immersive AI component in order to determine whether augmented and virtual reality (AR / VR) environments create opportunities for inclusive participation by newcomer communities (for example, enabling newcomers to create scenarios that accurately reflect their experiences rather than have scenarios created for them). The project will also create a framework to evaluate AI safety in migration contexts; and to determine if there are adequate guardrails currently in place.

BD-branding-line-blue - 1

Status

This project is in the proposal stage.

Expected start: 2026

BD-branding-line-blue - 1

Keywords

generative AI; large language models; diffusion models; AR/VR; immigration; migration discourse; algorithmic bias; AI safety; guardrails; interpretability; participatory design; civic participation; newcomer communities

Image Credits: Hanna Barakat  & Archival Images of AI + AIxDESIGN / https://betterimagesofai.org (external link)  / https://creativecommons.org/licenses/by/4.0/ (external link)