Human-Centered AI in Social Work: Who Gets Included and Who Gets Left Out?
“Human-centered AI” has become one of the most common phrases used in conversations about ethical technology. It appears in policy documents, organizational mission statements, and training materials across sectors, including social work. On the surface, the phrase sounds aligned with our professional values. After all, social work is a human-centered profession. But as social workers, we are trained to ask deeper questions—especially when language sounds universally benevolent. When we say human-centered AI, whose humanity are we centering? And just as importantly, whose humanity is being overlooked, marginalized, or erased?
Why Social Workers Must Interrogate “Human-Centered” Narratives
Social work has long understood that claims of neutrality often mask power. The profession emerged from an awareness that systems designed “for everyone” rarely serve everyone equally. Human-centered AI is no different. While the term suggests care, dignity, and inclusion, it often assumes a universal human experience that does not account for structural inequality, racism, ableism, colonialism, or economic exclusion. Without critical analysis, human-centered AI risks reproducing the same inequities social workers are ethically bound to challenge.
Human-centered AI frameworks are frequently developed by technology companies, policy leaders, and researchers situated in the Global North, often without meaningful input from the communities most impacted by algorithmic decision-making. When social workers adopt these frameworks without interrogation, we risk endorsing a version of “human-centered” that centers dominant identities while marginalizing others.
Human-Centered AI and the Limits of Individualism
One of the most significant tensions between social work values and mainstream human-centered AI frameworks is the emphasis on individual users rather than collective, relational, and structural realities. Many AI ethics models focus on user experience, consent, and individual harm, but they often fail to address systemic oppression embedded in data, design, and deployment. Social workers know that harm does not occur only at the individual level—it is produced and maintained through institutions, policies, and power structures.
When AI tools are evaluated primarily on usability or efficiency, the broader social consequences are often overlooked. Predictive analytics in child welfare, automated risk assessments in healthcare, and algorithmic screening in public benefits systems may be labeled “human-centered” because they aim to improve outcomes, yet they frequently intensify surveillance and control over marginalized communities. A truly human-centered approach must ask not only whether AI works, but who bears the cost when it does.
Anti-Oppressive Practice as a Lens for AI Ethics
Anti-oppressive social work practice requires more than harm reduction. It demands that we examine how power operates within systems and how those systems shape lived experience. Applying an anti-oppressive lens to AI means interrogating who defines the problem AI is meant to solve, whose knowledge is considered legitimate, and who is excluded from decision-making processes.
AI systems are trained on historical data that often reflects racialized, gendered, and class-based inequities. When these systems are framed as objective or neutral, they obscure the ways oppression becomes encoded into technology. Social workers must challenge narratives that position AI as merely a tool, rather than as a social product shaped by values, assumptions, and political priorities. Anti-oppressive AI practice requires naming bias explicitly and resisting the temptation to treat technological fixes as solutions to structural problems.
Who Is Missing from “Human-Centered” Conversations?
In many discussions of human-centered AI, the voices of service users, frontline workers, and communities experiencing systemic harm are notably absent. Individuals with disabilities, undocumented communities, Indigenous peoples, Black and Brown communities, and those living in poverty are often the subjects of AI systems rather than co-designers of them. Their experiences are reduced to data points, risk scores, or behavioral predictions rather than honored as sources of knowledge.
Social workers are uniquely positioned to disrupt this pattern. Our profession values participatory practice, community engagement, and lived experience as expertise. A human-centered AI framework that excludes these perspectives is incomplete at best and harmful at worst. Inclusion must go beyond consultation and move toward shared power in design, governance, and accountability.
Human-Centered AI Is Not the Same as Justice-Centered AI
One of the most important distinctions social workers can bring to AI conversations is the difference between human-centered and justice-centered approaches. Human-centered AI may prioritize comfort, convenience, or efficiency for users, while justice-centered AI prioritizes equity, accountability, and redistribution of power. These are not the same goals, and they sometimes conflict.
For example, an AI tool that speeds up case documentation may benefit practitioners but still reinforce surveillance practices that harm clients. A justice-centered approach would ask whether efficiency comes at the expense of dignity, privacy, or self-determination. Social work ethics require us to weigh these tensions carefully and resist solutions that prioritize institutional convenience over human rights.
What Social Workers Can Do Differently
Social workers do not need to become computer scientists to engage critically with AI. What we bring is ethical reasoning, systems thinking, and a commitment to social justice. In practice, this means asking difficult questions in classrooms, supervision, agencies, and policy spaces. Who designed this tool? What data was used? Who benefits from its use? Who is harmed or excluded? What assumptions about behavior, risk, or worth are embedded in the system?
It also means advocating for AI literacy that includes power analysis, not just technical skills. Social workers should be equipped to challenge harmful technologies, not simply adapt to them. This includes supporting students and practitioners to develop the language and confidence to say no to AI use when it conflicts with ethical practice.
Reclaiming Human-Centered AI Through Social Work Values
If human-centered AI is to mean anything in social work, it must be reclaimed through the profession’s core commitments to dignity, equity, and collective well-being. A social work–informed vision of human-centered AI would center relationality, cultural humility, and accountability. It would treat technology as a site of ethical struggle rather than inevitable progress. And it would insist that the people most impacted by AI systems have a meaningful role in shaping them.
Social workers have always challenged systems that claim to serve humanity while reproducing harm. AI is simply the latest system demanding our critical attention. The question is not whether social work belongs in conversations about AI ethics—it does. The real question is whether we are willing to bring our full critical, anti-oppressive lens into those conversations, even when it complicates dominant narratives.
The Bottom Line
Human-centered AI is not inherently ethical, inclusive, or just. Without critical analysis, it risks centering the same voices and values that have historically excluded marginalized communities. Social workers must move beyond surface-level ethics and engage AI as a site of power, resistance, and possibility. By grounding AI conversations in anti-oppressive practice and critical pedagogy, social work can help ensure that technology serves not just “humans” in the abstract, but real people in all their complexity.
The content in this blog was created with the assistance of Artificial Intelligence (AI) and reviewed and edited by Dr. Marina Badillo-Diaz to ensure accuracy, relevance, and integrity. Dr. Badillo-Diaz's expertise and insightful oversight have been incorporated to ensure the content in this blog meets the standards of professional social work practice.