Written by: Emmanuel Onyemaechi Ugwu, Furkan Gökalp, Giuliana Speranza, Yagmur Erdogmus
Session contributions and insights by: B.S
(Note: This blog post is a co-created reflection from VOICIFY’s second “Technology and Migrant Justice” Co-Creation Lab. The insights, quotes, and structural critiques presented below were generated collectively by the participants during our live session.)
Navigating words and worlds
We started the session with a prompt that helped us reflect on how we encounter technologies in our everyday lives as migrants in the EU.
For example, B.S indicated that “…they have both in the Czech Republic and Lithuania that, for certain governmental things, like taxes or health insurance… you can log in online with your citizen identity, and so you can do so many things, or, like, pay for your electricity, or, check your taxes, or your pension, check where your pension is at, and things like that. That would make it easier, as in you don’t need to go to an office anymore and stand in line. You can very easily do that online, but if you’re not a citizen, that option is not available to you, because you’re not recognized by the citizen portal, if I can call it like this. So it’s still, you know, not really making the immigrants’ lives better in that way.” She also mentioned the “too personal” information that is often required for submitting materials for renewing residence permits: “Do they actually need my siblings’ information?”
In contrast to the “too personal” questions B.S mentioned, Furkan’s point stressed the impersonal nature of automated procedures, such as AI chatbots, government systems and AI-based recruitment processes, which are supposed to be helpful and make things easier, but end up providing too generic answers: “it doesn’t make it somehow more reachable to you. You still need to go through those processes… You always need to either read the whole document of 100 pages, or you need to still talk with a human agent ”.
Giuli mentioned how the automated translations, while mostly helpful when you don’t know the language, can also be misleading and an obstacle when trying to learn the language. One good example is when trying to navigate the labor market in a country where you have jobs available depending on which languages you speak; automatic translation can make you think that the job’s main language is English until you realise the automated translation at work: “Not against the (automatic) translation, but it has also stopped my learning in a way, because you get so comfortable with just seeing in English everything”. Furthermore, Giuli explains how sometimes reading a job position in German online can be a good test of her level of the language and key to help her decide if to apply for it or not. However, automatic translations can give the illusion that some positions are available for English-speaking applicants when, in reality, they are not, acting as a barrier to her language learning and, in this case, with German. At the same time, she further noted how our focus on language is contingent on individual skills and the country we live in. For instance, if we were living in the UK, perhaps, language navigation would not come to our attention in thinking about our encounters with technologies as migrants. This suggests that the intersection of digital technology and migration transcends individual experiences to include contexts, systems, and sociocultural realities in destination countries.
Emmanuel echoed this ambivalent sentiment and noted the fluctuation of his mood with regards to the “ease” digital technologies provide and feelings of “proving your worthiness”: “Why can’t Germany consider that many people are immigrants and add English versions to some letters?”. At the same time, he further noted how this focus on language adaptation is also indicative of a certain privileged position, rather than, for example, the precarity of being undocumented while navigating the digital systems. His expression further highlighted the psychological consequences of constantly relying on digital technology. For instance, reading everything or shopping with Google Translate diminishes one’s sense of self and may result in social anxiety over time. This could deter social integration due to uncertainties about how others may perceive you.
Mitigating bias, creating bias: Dialect Recognition and Casematcher
Following Emmanuel’s point on how technologies matter at the level of documentation for migrants, we reviewed two AI technologies used in asylum systems: Dialect Recognition, used in Germany (Dumbrava, 2025), and Casematcher, used in the Netherlands (Memon et al., 2024). We reflected on the main objectives of these technologies, their prospects, inherent challenges, and whose right it is to address such challenges. We collected responses to the individual questions as follows.
What problem is technology trying to solve? What might it make easier/better?
What could go wrong? What happens to the person if the system gets it wrong?
Who has the power to correct the system?
What would make this use of AI more just?
One of the central challenges raised during the workshop was how genuinely difficult it is to make AI usage more just, particularly in a context with so many competing stakeholders. Oweke (2026) writes that the question of AI governance is key since more AI systems are being developed and deployed across borders with no accountability to the populations they affect, and how many countries are “racing to establish domestic AI governance mechanisms, often in distinct ways” to protect their agency and sovereignty in the face of a transformative transnational technology. In the session, Giuli also highlighted how this multiplicity of actors makes it especially hard to define who holds the power to define the borders of AI usage. Relatedly, this concern connected to a broader call for stronger human oversight. The particular emphasis was put on ensuring that the people directly affected by the asylum process are not sidelined by automated systems but remain central to decisions that shape their lives. Furkan added that this requires a clearly defined code of benefits that explicitly prioritises the person over technical convenience, and he argued that addressing this challenge properly demands something more fundamental: a broader reflection on what it means to be human in an age increasingly mediated by AI.
These discussions helped us situate the possible individual and societal consequences of the use of these (digital) technologies in migration governance: not to villainize nor to glorify their use and power, but rather to understand how they might implicate livelihoods at different levels. This led us to reflect on how we understand humans and cultures, as both Emmanuel and Furkan reflected on whether a digital tool can “empathize” or know the nuances of a dialect, language, story or culture. As Giuli noted, who holds the power when there are too many stakeholders, and what happens to accountability?
References
Dumbrava, C. (2025). Artificial intelligence in asylum procedures in the EU. Think Tank | European Parliament. https://www.europarl.europa.eu/thinktank/en/document/EPRS_BRI(2025)775861
Oweke, T. (2026). Who is Accountable When AI Goes Global?. The Council on Foreign Relations. https://www.cfr.org/articles/who-is-accountable-when-ai-goes-global
Memon A, Given-Wilson Z, Ozkul D, et al. Artificial Intelligence (AI) in the asylum system. Medicine, Science and the Law. 2024;64(2):87-90. doi:10.1177/00258024241227721
