AI for Victim Justice in Refugee Status Determination: Enhancing Fairness and Protection

Introduction

The primary and universal definition of a refugee has been provided in the 1951 Refugee Convention, defining refugee as someone who, owing to well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable or, owing to such fear, is unwilling to avail himself of the protection of that country; or who, not having a nationality and being outside the country of his former habitual residence, is unable or, owing to such fear, is unwilling to return to it (UNHCR, n.d.). Every person who seeks international protection as a refugee has to undergo the process of refugee status determination (RSD).

The objective behind this process is to figure out the reason why a certain person cannot return to their origin country. Taking in the context of international and humanitarian principles, the definition of refugees plays a very major role in determining the status and entitlements of individuals fleeing persecution and seeking asylum in different foreign countries.

This definition forms the bedrock of international refugee protection as it not only answers ‘who is a refugee’ but also determines their entitlements and rights. The principle of non-refoulement which ensures that refugees are shielded from returning to their country of origin, where they may face persecution or serious harm. The declaratory nature of the refugee definition emphasizes the immediate recognition of refugee status once the criteria are met, safeguarding individuals during the process of RSD. The RSD process involves evaluating an individual’s circumstances against the criteria set forth in the refugee definition to establish their eligibility for international protection. Therefore, it becomes essential to approach RSD with sensitivity and efficiency, recognizing the vulnerabilities and urgent needs of individuals seeking refuge.

This is where artificial intelligence (AI) comes into the picture, as it has the potential to address these challenges. AI can be used to automate tasks in RSD, such as the analysis of refugee testimonies. By using AI, human decision-makers will get more room to focus on more complex cases. AI will certainly help to identify patterns and trends in refugee cases that would be difficult to detect by human decision-makers. AI can be used as a tool to ensure that refugees who are genuinely in need of protection are not denied asylum. Consequently, the integration of AI into RSD, either as a complementary or alternative method to human credibility assessments, holds the potential to improve efficiency and accuracy.

The integration of AI into RSD holds immense potential to transform the assessment of refugee claims, ensuring accuracy and efficiency in the decision-making process. By addressing the deficiencies in credibility assessments and streamlining the recognition of refugee status, AI can significantly contribute to safeguarding the rights and protection of those in need of asylum. However, the responsible implementation of AI in RSD requires a thorough understanding of its capabilities and limitations, accompanied by proactive measures to prevent bias and ensure the fair treatment of refugees.

The second part of the article delves into the increasing domain of AI as a tool for humanitarian aid and categorically debates how the use of AI could create a much better medium to give aid to those in urgent need with the help of examples. Later, the authors argue that AI paves a much better way to provide a means for assessment of refugees in their status determination, as it can serve as a means free from all human biases and complexities. However, there are certain limits to the use of AI, but those can be eliminated with gradual use and experience gained over time.

The third part evaluates various factors that could affect the various aspects of RSD. Credibility assessment is the major facet of RSD, and it is affected by the elements of subjectiveness, which could severely affect human emotion and a wrong interpretation of those could lead to major flaws in the process of credibility assessment. This emotion comes forward while the claimant has to describe various facts to the decision-maker on which they could come to a conclusion regarding their status in which AI could do a much better job, as it could form a decision based on those standards of material facts which could be subjected to human error and bias. The last part evaluates the merits and demerits of the contentions explained.

Artificial Intelligence: As Humanitarian Tech

One fascinating truth about technology is that it can address problems even outside the technology domain by providing practical solutions to complex sets of problems and orchestrated workflows. AI has become a buzzword within the humanitarian sector in recent years. Much like ‘blockchain’ or ‘drones’, it is an area where new technology is developing quickly and operators are keen to test its possible applications. From an economic perspective, it is a big business too: from a total of $1.3 billion raised in 2010 to over $40.4 billion in 2018, funding has increased at an average annual growth rate of over 48 per cent. (Gueuning, 2020).

Regardless of how complex a technology might be, a humanitarian tech must be developed with the ultimate goal of helping those in need.

The idea of ‘humanitarian design’ is about creating such products or services that can help populations affected by natural or man-made disasters. The primary goal must be to make a positive impact with a determination to make a change so that a crisis can be averted with minimal suffering and damage. Humanitarian tech could be used to address issues related to vulnerable populations, such as rehabilitation, refugee living conditions and their health, and settlement as these techs can significantly reduce the time taken to address such issues by reducing intermediaries’ participation with inherent biases and power politics (Rajani et al., 2014). AI holds significant potential in the humanitarian sector, offering predictive analytics for early crisis preparation (AI-driven models can anticipate and forecast migration patterns during conflicts, famines or natural disasters), image recognition for disaster zone identification (can be used to assess the extent and impact of disaster quick and accurately), natural language processing for effective communication, and adaptive web design for personalized user interactions (so that user receives the most relevant information made as in accordance of their needs). Implementing these AI-driven solutions can lead to more efficient and targeted humanitarian responses, ultimately benefiting the communities affected by crises and disasters. These techs can enable the subject population to share information and communicate directly instead of being dependent upon some intermediaries.

The use of any new technology in humanitarian settings can have positive as well as negative consequences to bear. On the one hand, this technology can be used to improve the lives of numerous vulnerable populations through improving communication and coordination between various aid organizations to help in providing appropriate care, food and shelter, and on the other hand, it can be used on these isolated refugees as a test subject for collection of data so that these techs can be improved further with the goal of making it safer for further use (Fleischmann, 2000). With the use of these technologies, it also becomes important to ensure that such use must be ethical and the data collected in this process should be protected and used responsibly as it would be disastrous if such sensitive data were revealed or leaked. Hence, it is essential to critically assess and evaluate the potential risks and unintended consequences of deploying technology in humanitarian contexts. This evaluation should always be conducted in consultation with the affected population so that it becomes easy to identify and mitigate any negative effects that may arise out of using such tech. To smooth out the implementation of AI in humanitarian contexts such organizations and NGOs also need to train and educate their own personnel on such technologies and to achieve this there must be a transfer of knowledge between the experts and those forming the workforce enabling all members to understand how to use these techs to get the most out of it for a certain problem (Gueuning, 2020).

The use of humanitarian tech is a complex issue with no easy and immediate answers, but it is also important to weigh the potential benefits of using this technology against the potential risks. For example, if a tech is designed only from a single perspective and ignores all the other local contexts and needs of the affected population, it can lead to unintended consequences, as ultimately it can transfer risk to them only, whom it intended to help. It should be used in a way that does not strengthen existing inequalities and discrimination against affected persons. This approach ensures that the unique perspectives, diverse needs and individuals’ knowledge are taken into account, which moves beyond Western-centric views and allows room for appropriate solutions that are culturally sensitive. By creating technology infused with the concerns of the affected population, there becomes a greater chance of developing solutions that will truly meet their requirements and respect (Jacobsen & Sandvik, 2018). Empowering the community in the design process of such humanitarian tech fosters a sense of ownership and agency over the technology, promoting sustainable and locally embedded solutions. By doing so, technology can be harnessed as a force for positive change, empowering vulnerable populations and contributing to more equitable and resilient humanitarian responses.

The current parameter of credibility assessments for RSD poses significant problems, and it proves to be a complex challenge for the government as they would have to undergo in-depth transformation while also meeting their relevant constraints. These assessments, which determine the authenticity of an asylum seeker’s claims, are varied, inefficient and highly uncertain due to the subjective nature of fear that refugees which they must demonstrate. Such subjectivity often leads to inconsistent outcomes and raises questions about the fairness and feasibility of the current credibility assessment process. This becomes a problem because it means that not all refugees who are eligible for protection will be granted it, and the process can be long and stressful for refugees, who may have to wait months or even years for a decision.

Governments and international organizations, including the United Nations High Commissioner for Refugees (UNHCR), have recognized the deficiencies in credibility assessments and the pressing need to address them.

As a result, the integration of AI into RSD has gained attention as a means to overcome these challenges and better manage the heavy caseloads faced by authorities. Furthermore, the widespread adoption of AI in various government programmes, particularly those related to migration and border control, as well as its growing utilization within the United Nations system, has paved the way for exploring its application in RSD.

Terming the new system as better than the previous one would be an oversimplification of the underlying problem (Hathaway, 2021). To fully understand the implication of AI in RSD, it is pertinent to analyse the development AI will bring in decision-making and better organized aid delivery and its effects on refugees themselves. Before going in awe of the positive outcomes that AI will bring in RSD, it is pertinent to evaluate the limitations of the existing system so that those gaps could be identified which the technology might be able to fill. If we focus on the deficiencies and shortcomings of the current credibility assessment system, it will enable us to better assess how AI should address these loopholes and contribute to faster and fair decision-making. This socio-legal approach ensures that the implementation of AI in RSD is carried out with a clear understanding of its specific purposes and expected impact on refugees’ lives.

Human Problem of Credibility Assessments and Subjective Fear

One of the major things that affect the credibility assessment is the availability of the material fact it refers to, the facts that are needed by the person to establish to the decision-maker that the fear or hardship he is facing is real. Further, these facts must be based on evidence, like his own oral testimony or expert reports and information received from his own country (this will establish well-founded fear by the claimant). This process of credibility assessment is solely based on finding out whether the person claiming refuge has a justifiable cause or not, and for this, the person in charge of arriving at a decision needs relevant information (provided by the claimant which forms the basis of subjectivity), which is then verified with all that information already available (from different sources which forms the basis of objectivity). Only after this process, it could be determined whether that information given by the claimant is relevant or not. The fear of persecution in their own country is one of the most crucial factors that need to be determined through credibility assessment before a person can be given refugee status (UNHCR, n.d.). However, this criterion has been criticized as many authors view it as a means to increase complexity as in this appropriate consideration has to be given to subjective fear and also a situation to support that objective validation of that fear (Hathaway, 2021).

The concept of ‘well-founded fear’ as a standard for RSD has been criticized on many occasions with respect to the presence of subjective elements. Some advocate for a purely objective inquiry in the process of RSD while others lean towards the importance of subjective fear as a criterion for RSD by contending that if a claimant is not perceived as subjectively fearful even with the presence of objective elements, he could be denied refugee status. This brings us to one of the major challenges in decision-making, which is the lack of standardized criteria which could help decision-makers in assessing credibility and it potentially leads to biased decisions and inaccuracies.

The federal court of Canada in the case of Yusuf v. Canada expressed concerns at this well-founded fear standard by stating that:

I find it hard to see in what circumstances it could be said that a person … could be right in fearing persecution and still be rejected because it is said that fear does not actually exist in his conscience. The definition of a refugee is certainly not designed to exclude brave or simply stupid persons in favor of those who are more timid or more intelligent. Moreover, I am loath to believe that a refugee status claim could be dismissed solely on the ground that as the claimant was a young child or a person suffering from a mental disability, he or she was incapable of experiencing fear, the reasons for which clearly exist in objective terms. (Yusuf v. Canada (Minister of Employment and Immigration, 1991)

Amidst all this, the role of the decision-maker becomes much crucial in RSD as the lack of standard assessment criteria leaves wide discretion in the hand of decision-makers as they must evaluate a claimant’s plausibility primarily based on their own testimony; often there is not even any concrete evidence beyond it. At that time, decision-makers have to rely on their own judgement and personal experience which increase room for error in the process.

The use of credibility indicators is a step in the right direction, but it is not enough to ensure a fair and objective assessment process as they do not fully address the subjective elements in the ‘well-founded fear’ standard. This means that decision-makers’ personal judgements and perceptions can still play a significant role in the assessment process, which could lead to potential bias and inaccuracies. Much efforts are needed to standardize and improve the assessment criteria for refugees seeking protection worldwide. This could involve developing more objective criteria for assessing credibility, as well as providing decision-makers with more training on how to use these criteria fairly and consistently.

The use of objective criteria in credibility indicators aims to avoid reliance on ‘instinct’ by considering consistency, coherence and corroborative evidence. However, this objectivity can allow decision-makers to speculate about an asylum seeker’s actions without considering relevant compounding factors. The indicators may lead to speculative assessments of cognition, demeanour and risk response, adding unnecessary burden to already vulnerable claimants and complicating an already uncertain decision-making process. Credibility indicators, despite being folded into various criteria, perpetuate the challenges of assessing subjective fearfulness in refugee credibility evaluations.

Conclusions

Due to the increase in inflation in the economy, the government is bound to cut down its expenditure but not at the cost of welfare administration. Therefore, to maintain administration at a reduced cost and to deliver administration at a speedy and proper time, the government has switched to AI tools. The use of AI mechanisms in this area has accelerated the application process of RSD. The above analysis of the opportunities and risks of using AI in RSD recognizes new technologies can create risks for people with limited agency. In simple terms, as AI continues to advance, policymakers and legal experts need to work together to create a framework that combines technology with humanitarian principles to help those seeking international protection. In the next part of this paper, we will explore how AI can be used as a tool for humanitarian aid, making it easier to assist people in urgent need. AI can also be beneficial in assessing refugee status, as it can be free from human biases and complexities. While there are some limits to using AI, with gradual use and time, we can overcome these challenges.

Nowadays AI is used in all sectors of the economy and it has positively affected the efficacy. Therefore, introducing AI in this sector will change the lives of miserable populations seeking RSD in the entire globe. In all challenging situations with proper advancement and development, technology improves the lives of human beings. Without acknowledging the potential benefits and harm signal to AI in RSD cannot be given. There is a need to bifurcate the advantages and disadvantages of AI for refugees. By understanding various aspects of this, the government can work to remove disadvantages and ambiguity and maximize the advantages. The analysis of data and identification of the problem and issues in a system can be easily done by AI tools. The problem lies in the decision-making of humans based on the statements and information analysed by AI tools. This shows that AI can be used to identify problems but without human interventions, the identified problems and issues cannot be fixed. Ultimately, harnessing technology with a human-centric approach empowers vulnerable populations and enhances humanitarian responses for a more equitable and resilient world. Also, this assessment can be challenging because it depends on subjective elements like personal experiences and emotions. Decision-makers often rely on their judgement, leading to potential biases and inaccuracies.

In simple terms, credibility assessment in RSD is essential to determine if a person qualifies for refugee protection. It involves evaluating the truthfulness of the claimant’s fear and hardship. However, this cannot be assumed that AI is weak. AI brings into notice the faults in the course of appraising the integrity of the refugee. The present standard for determination is very subjective and open to interpretation. This brings lots of human interventions that cause biases and inaccuracies. AI provides an objective and standardized approach to assessment using proper algorithms prepared by experts, and this ultimately creates a benchmark. Using technology like AI ensures objectivity and it also minimizes the subjective element. This ensures consistent and fair decisions in granting refugee protection and this makes the process transparent. AI allows assessment in a fast manner, precisely in situations where there is a clear and widespread threat to individuals in a particular region or country. However, there is a need to balance between simplified assessment and individualized assessment that considers subjective fear. Both approaches have their benefits, but lucid and effective RSD should include both elements, based on the specific circumstances and needs of asylum seekers. In some cases, it is not feasible to perform individualized assessments as there is an influx of refugees on a large scale; therefore, refugee status is recognized based on readily apparent and objective circumstances in the country of origin. In situations of mass displacement with large numbers of people seeking refuge, individual assessments can be cumbersome and lukewarm; therefore, prima facie determinations offer a more efficient approach by granting refugee status to a group of asylum seekers based on common and visible conditions in their home country that indicate a high likelihood of persecution or danger. These objective circumstances may include factors such as armed conflict, widespread violence or severe human rights violations in the country of origin. The idea is that if conditions are so grave and widely known, it becomes clear that a large number of people fleeing from that country are likely to be refugees in need of international protection. Accelerated case processing is a strategy used in countries by the government of that country to expedite the RSD process for certain cases. This approach is typically applied to cases where the claimant’s need for protection is readily apparent, and their eligibility can be determined quickly without the need for a lengthy and detailed examination. In accelerated case processing, the focus is on objective criteria, such as country of origin information, prevailing conditions in the country, and other readily available evidence. It is often used in situations where there is already substantial evidence or knowledge about the claimant’s country of origin, indicating a high likelihood of persecution or danger. This strategy allows for swift decision-making and faster access to protection for those whose claims can be easily verified and established based on objective information. However, it is crucial to ensure that this approach does not compromise fairness and accuracy in determining refugee status. Therefore, a balance must be struck between expedited processing and ensuring that individuals with valid claims receive the protection they need. Both temporary protection arrangements and suspension of RSD processing are responses to urgent and large-scale humanitarian crises, where individualized assessments may not be feasible or necessary.

However, it is essential to regularly review and reassess these measures to ensure that they remain fair, effective and consistent with international refugee protection principles.

Indeed, in the era of the current ‘AI spring’ where advancements in AI are reshaping various sectors, it is crucial to reevaluate the necessity of requiring asylum seekers to ‘prove’ subjective fear in the RSD process. With the potential integration of AI in the RSD process, it opens up new possibilities to address this issue. AI can analyse broader patterns, trends and country conditions to assess the legitimacy of asylum claims objectively. It can use data-driven analysis to identify patterns of persecution or danger in certain regions, making the process fairer and more efficient. By reducing the emphasis on subjective fear, the RSD process can become more equitable and lessen the burden on vulnerable asylum seekers. The use of AI can lead to more accurate and consistent decision-making, ensuring that those genuinely in need of international protection are not unfairly denied asylum due to the challenges of proving subjective fear.

However, it is essential to proceed with caution and ensure that the implementation of AI in RSD remains ethical, transparent and respectful of human rights. The goal should be to strike a balance between objective analysis and individualized assessments, making the process more humane and just for those seeking refuge. As technology continues to advance, we must seize the opportunity to reform the RSD process to better serve the needs of asylum seekers and uphold humanitarian principles.