22 10.1163/22131035-14020001 | enigbokan decision.119 However, this approach has been criticised for their low fidelity and for disallowing an inspection and verification of a black-box system’s rationale.120 Developers of these systems stress that the explanations have no direct connection between the features that cause the black-box system to make its decisions and those mentioned in the justification. Explainable ai systems have to infer, i.e indirectly work out, the black-box model’s rationale.121 Therefore, the ai explanation can be an inaccurate representation of the original (black box).122 Since the black-box model’s processing is opaque, people will not be able to check the explanation’s accuracy. This makes the trustworthiness of such ai explanation questionable.123 Therefore, completely replacing human decision-makers with ai algorithms in the determination of statelessness risks diminishing transparency and violating the administrative duty to uphold it. Transparency emphasises the need for explainable ai, where the logic and rationale behind ai-driven decisions are made clear.124 It becomes difficult to challenge an algorithm decision without a clear explanation of the decision.125 Van den Hoven relates this challenge to hermeneutical injustice.126 A ‘condition where an individual is increasingly ‘dispossessed of the interpretive tools, concepts and even words to make sense of the world and of one’s experiences’.127 Machine learning’s opacity may make its owners less accountable, which would undermine public trust, and make their decisions incontestable.128 Individuals must be given reasons behind decisions that impact their rights or interests so they can challenge decisions they believe to be unfair and make the required representations during the process.129 Therefore, the absence of reasoned decisions undermines the individual’s ability to challenge human rights 119 120 121 122 123 124 125 126 127 128 129 Ibid, 974–975. Ibid, 965. Ibid. Ibid. Ibid. Khaled Khalaf Abed Rabbo Aldrou, (n 105) at 4. Antoni Roig, ‘Safeguards for the Right not to be Subject to a Decision Based Solely on Automated Processing’ (Article 22 gdpr)’ (2017) 8 European Journal of Law and Tech 3. Emilie van den Hoven, ‘Hermeneutical injustice and the computational turn in law’ (2021) 1 Journal of Cross-Disciplinary Research in Computational Law 2; Miranda Fricker, Epistemic Injustice: Power & the Ethics of Knowing(Oxford University Press, 2007) 158. Ibid. Mireille Hildebrandt, ‘The New Imbroglio Living with Machine Algorithms’ (2016) in Liisa Janssens (Eds.), The Art of Ethics in the Information Society: Mind You (Amsterdam University Press, 2017) 55–60. Handbook on Statelessness supra n 13 Para 71;art. 31 1954 Convention. South Africa’s Promotion of Administrative Justice Act 3 of 2000. International Human Rights Law Review (2025) 1–31