Independent and cutting-edge analysis on global affairs

Artificial Intelligence (AI) can be defined as “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”[1] The application of AI in asylum decision-making and migration governance is a new concept experimented by only a handful of states. For instance, while Canada uses algorithmic decision-making in immigration and asylum determination[2], the German Federal Office for Migration and Refugees (BAMF) uses a pilot program, which identifies different dialects in Arabic by using voice samples, in order to verify the nationality of the asylum seeker.[3] The United States (US), on the other hand, has been using a Risk Classification Assessment System, that uses deep learning, which recommends immigration officers whether a migrant should be detained or released on bail.[4] One of the areas where AI and emerging technologies have been utilized is resettlement. 

What Is Resettlement?

Resettlement is defined by the UN Refugee Agency (UNHCR) as: “the selection and transfer of refugees from a State in which they have sought protection to a third State which has agreed to admit them – as refugees – with permanent residence status.”[5] The Convention relating to the Status of Refugees[6] (Refugee Convention), which is the international convention governing the status and rights of refugees, is largely silent on the issue of resettlement. As a result, states have the sovereign competence to decide if, who and how many persons to resettle. This leads to divergencies between national resettlement practices with regard to the selection criteria; resettlement procedures and the status granted to resettled persons.

Today 31 states including the US, UK, Canada, Sweden, Norway and Germany engage with resettlement around the world.[7] UNHCR usually acts as an intermediary in resettlement and facilitates the resettlement of refugees under UNHCR’s mandate for whom resettlement is the most appropriate durable solution, and who fall under UNHCR’s Resettlement Submission Categories. These categories include “Legal and/or Physical Protection Needs, Survivors of Torture and/or Violence, Medical Needs, Women and Girls at Risk, Family Reunification, Children and Adolescents at Risk, and Lack of Foreseeable Alternative Durable Solutions”.Refugees resettled under the auspices of UNHCR usually receive permanent residency and have access to civil, political, economic, social and cultural rights similar to those enjoyed by nationals.-

Resettlement classically targets refugees and in particular vulnerable individuals including victims of torture, women and girls at risk, minors, elderly and those in need of medical care.

Resettlement classically targets refugees and particularly vulnerable individuals including victims of torture, women and girls at risk, minors, the elderly and those in need of medical care. Whilst some states simply rely on UNHCR for referrals or conduct selection missions on the basis of UNHCR referrals, others conduct their own resettlement selection missions in third countries with the participation of national migration authorities and sometimes non-governmental organizations. Similar to the selection criteria and procedures, the rights granted to resettled refugees as well as the scale and size of the resettlement initiatives also differ significantly between states. 

Application of AI and Emerging Technologies in the Context of resettlement

The idea to use AI and emerging technologies in the context of resettlement is not entirely new. In 2018, Bansak et al. came up with an algorithm that uses a combination of machine learning and optimal matching to assign refugees to certain locations in the resettlement state where they are more likely to find employment and have a better chance for integration.[8] Bansak et al. argue that their proposed algorithm would lead to gains of “roughly 40 to 70 percent, on average, in refugees’ employment outcomes relative to current assignment practices.” Building on this proposal, Jones and Teytelboym in a 2018 article proposed a centralized matching system that takes into account the preferences of refugees as well as local communities in determining where exactly refugees would be resettled.[9] It is reported that the software ‘Annie’[10] developed by a number of scientists in the United Kingdom, the US and Sweden, which assigns refugees to regions (areas) in the resettlement state where they would have better integration, is applied in the US as a pilot project and has yielded quite positive results.[11] The Swiss government uses similar software developed by scientists from the University of Stanford to distribute resettled refugees to different cantons in Switzerland.[12] Although it is estimated that resettled refugees are more likely to be employed when they are resettled by using the said algorithms, Molnar and Gill rightly point out that the use of such systems may also exacerbate inequalities by placing refugees with the least prospect of success into under-resourced areas.[13] In addition to the algorithms outlined above, chatbot software, triage applications as well as web and mobile applications, are already being used in different states, to facilitate resettlement and integration of refugees, resettled or otherwise, to the host societies.[14]

What Are the Risks and Why Do We Need Further Research on it?

While the use of mentioned pilot software and programs which aim to match refugees with locations in the resettlement state where they have a better integration prospect does not present a major human rights problem, the expanded use of AI in resettlement may do so. Potential use of AI and emerging technologies in resettlement processes includes conducting eligibility assessment for resettlement which usually compromises of refugee status determination and vulnerability assessment, triaging resettlement cases and referrals, matching refugees with resettlement states and integration fostering based on predetermined criteria and datasets. This kind of expanded use of AI in the context of resettlement can create serious issues in terms of human rights and interfere with, inter alia, the right to be free from discrimination, the right to privacy, right of due process of law and the right to an effective remedy. While the existing and potential use of AI in resettlement, needs to be scrutinized from a legal and social point of view, there is no in-depth study examining the legal and social implications of the use of AI and emerging technologies in the resettlement context although there are few recent studies and reports discussing the use of AI in different migration contexts and potential implications of such use.[15]

Potential use of AI and emerging technologies in resettlement processes includes conducting eligibility assessment for resettlement which usually compromises of refugee status determination and vulnerability assessment, triaging resettlement cases and referrals, matching refugees with resettlement states and integration fostering based on predetermined criteria and datasets.

Concluding Remarks

Whilst the use of artificial intelligence can support beneficial outcomes it can also bring new risks or negative consequences for refugees and society.[16] Thus, the existing and potential expanded use of AI and emerging technologies in the resettlement context needs to be scrutinized from a legal and social point of view and we need further interdisciplinary research on the application of AI and new technologies in the resettlement context.  Research is needed not just to identify benefits, risks, and the negative consequences of the application of AI and emerging technologies in the context of resettlement, but also to better understand its potential interference with fundamental rights. Resettlement may mean the difference between life and death of a vulnerable refugee such as a torture survivor, a girl at risk, or a seriously ill refugee, and leaving the destiny of vulnerable refugees to so AI without conducting in-depth research first seems plausible.


[1] The Oxford Dictionary of Phrase and Fable (2 ed., OUP 2006), available at: https://www.oxfordreference.com/view/10.1093/acref/9780198609810.001.0001/acref-9780198609810-e-423

[2] Petra Molnar and Lex Gill, Bots at the Gate: A Human Rights Analysis of Automated Decision-Making in Canada’s Immigration and Refugee System (University of Toronto, 2018), available at: https://citizenlab.ca/wp-content/uploads/2018/09/IHRP-Automated-Systems-Report-Web-V2.pdf

[3] Anna Beduschi, “International migration management in the age of artificial intelligence,” Migration Studies (2020), p. 1-21. 

[4] Estefania Mccarroll, “Weapons of Mass Deportation: Big Data and Automated Decision-Making Systemsin Immigration Law”, Georgetown Immigration Law Journal, Vol. 34 (2019), p. 717.

[5]UNHCR, Resettlement Handbook, (2011), available at: https://www.unhcr.org/46f7c0ee2.pdf

[6] Convention relating to the Status of Refugees, 189 U.N.T.S. 150, entered into force 22 April 1954.

[7] UNHCR Projected Global Resettlement Needs, (2021), https://reliefweb.int/report/world/unhcr-projected-global-resettlement-needs-2021

[8] Kirk Bansak, Jeremy Ferwerda, Jens Hainmueller, Andrea Dillon, Dominik Hangartner, Duncan Lawrence, Jeremy Weinstein, “Improving Refugee Integration through Data-Driven Algorithmic Assignment,” Science (2018): p. 325–329. 

[9] Will Jones and Alexander Teytelboym, “The Local Refugee Match: Aligning Refugees’ Preferences with the Capacities and Priorities of Localities,” Journal of Refugee Studies, Vol. 31, No. 2 (2018): p. 152–178.

[10] See for instance Annie™ MOORE (Matching and Outcome Optimization for Refugee Empowerment) which is presented as the world’s first software that helps resettlement agencies optimize their initial placement of refugees within host countries. Cf. Refugee AI website, available at: https://www.refugees.ai/

[11] Andrew C. Trapp, Alexander Teytelboym, Alessandro Martinello, Tommy Andersson, NargesAhani, 
“Placement Optimization for Refugee Resettlement,” Lund University Department of Economics Working Paper 2018:23, available at: https://project.nek.lu.se/publications/workpap/papers/wp18_23.pdf

[12]Petra Molnar and Lex Gill, (2018), p. 50.

[13]Petra Molnar and Lex Gill, (2018), p. 50.

[14] UNHCR, Chatbots in humanitarian settings: revolutionary, a fad or something in-between? (2021) available at: https://www.unhcr.org/innovation/chatbots-in-humanitarian-settings-revolutionary-a-fad-or-something-inbetween

[15]Beduschi (n 14); Molnar and Gill (n 13); Mccarroll (n 15); Emre E. Korkmaz (edn), Digital Identity, Virtual Borders and Social Media: A Panacea for Migration (Edward Elgar, 2021); Petra Molnar, “Technology on the margins: AI and global migration management from a human rights perspective,” Cambridge International Law Journal, Vol. 8, No. 2 (2019): p. 305-330.

[16] European Commission, “Opportunities and Challenges for the Use of Artificial Intelligence in Border Control,” Migration and Security, f Volume 1: Main Report (2020) and Volume 2, Addendum 21.4.202 COM (2021).

CONTRIBUTOR
Meltem Ineli-Ciger
Meltem Ineli-Ciger

Dr Meltem Ineli-Ciger is an Assistant Professor at the Suleyman Demirel University Faculty of Law in Turkey.

The Premium Corporate Sponsor of TPQ
Yapı Kredi
Foreword After the violent dissolution of the former Yugoslavia in the 1990s, there had been a shared sense of hope for a more peaceful future for the European continent. Unfortunately, this comfortability disappeared after Russian President Vladimir Putin ordered his troops to march against the Ukrainian forces throughout the border on 24 February 2022. This marked a turning point not only for the region...
STAY CONNECTED
SIGN UP FOR NEWSLETTER
FACEBOOK
PARTNERS