Scholarlink

<p>Artificial intelligence integration into healthcare represents not merely a technical advancement, but a critical juncture in the politics of care delivery. This narrative review interrogates how AI systems, when designed without attention to power asymmetries and epistemic injustice, risk encoding historical inequities into automated decision-making processes that disproportionately harm marginalized communities. Medical errors remain a persistent threat to patient safety, yet their distribution across populations reveals systemic patterns of violence rooted in institutional racism, linguistic exclusion, and digital colonialism. As AI becomes embedded in clinical workflows, the question is not whether these tools can optimize efficiency, but whether they will preserve human dignity and redistribute power in healthcare encounters. Drawing on Critical Race Theory and frameworks of Algorithmic Justice, this review critically examines how AI tools either reinforce or challenge existing hierarchies through five interconnected mechanisms: data sovereignty violations, algorithmic violence, opacity as corporate strategy, techno-solutionism that erodes human agency, and the digital colonialism embedded in Global North safety models imposed on Global South contexts. Through thematic analysis of current research, policy reports, and illustrative case studies, we demonstrate how AI-driven safety interventions often fail marginalized populations not despite their design, but because of it. The review reveals that algorithmic bias is not a technical glitch, but a manifestation of deeper societal power structures, while the “black box” problem reflects deliberate corporate secrecy rather than mere complexity. Ultimately, we argue for a paradigm shift: moving from AI deployment as technical implementation toward participatory governance models that center community accountability, recognize the rational basis of institutional distrust among historically surveilled populations, and preserve the doctor–patient relationship as an essential safeguard for the vulnerable. This requires reimagining patient safety not through optimization metrics, but through frameworks of justice, dignity, and collective liberation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automating inequity: how artificial intelligence reproduces systemic failures in patient safety for marginalized communities

  • 期刊论文

  • 发表日期：

    2026年2月25日

查看全文
  • Titus Oloruntoba Ebo,
  • Ayodele Osunmakinde,
  • Akinsola J. Asaolu,
  • Dolapo Mary Ebo,
  • Eghosasere Egbon,
  • David Bamidele Olawade

摘要

Artificial intelligence integration into healthcare represents not merely a technical advancement, but a critical juncture in the politics of care delivery. This narrative review interrogates how AI systems, when designed without attention to power asymmetries and epistemic injustice, risk encoding historical inequities into automated decision-making processes that disproportionately harm marginalized communities. Medical errors remain a persistent threat to patient safety, yet their distribution across populations reveals systemic patterns of violence rooted in institutional racism, linguistic exclusion, and digital colonialism. As AI becomes embedded in clinical workflows, the question is not whether these tools can optimize efficiency, but whether they will preserve human dignity and redistribute power in healthcare encounters. Drawing on Critical Race Theory and frameworks of Algorithmic Justice, this review critically examines how AI tools either reinforce or challenge existing hierarchies through five interconnected mechanisms: data sovereignty violations, algorithmic violence, opacity as corporate strategy, techno-solutionism that erodes human agency, and the digital colonialism embedded in Global North safety models imposed on Global South contexts. Through thematic analysis of current research, policy reports, and illustrative case studies, we demonstrate how AI-driven safety interventions often fail marginalized populations not despite their design, but because of it. The review reveals that algorithmic bias is not a technical glitch, but a manifestation of deeper societal power structures, while the “black box” problem reflects deliberate corporate secrecy rather than mere complexity. Ultimately, we argue for a paradigm shift: moving from AI deployment as technical implementation toward participatory governance models that center community accountability, recognize the rational basis of institutional distrust among historically surveilled populations, and preserve the doctor–patient relationship as an essential safeguard for the vulnerable. This requires reimagining patient safety not through optimization metrics, but through frameworks of justice, dignity, and collective liberation.