Artificial Intelligence has slipped into our daily routines almost without us noticing. It summarises long texts, translates emails, suggests wording, and sometimes even drafts entire documents. For many people, this feels convenient. But for social workers, youth workers, educators, and others who work with stories and relationships, AI raises a different set of questions about care, dignity and responsibility.
This blog explores why AI is not just “another tool,” and why the social sector in particular needs to approach it with intention and caution.
AI isn’t neutral, it reflects the world it learns from
AI systems learn by analysing huge amounts of human-made text. And that text comes from a world shaped by:
- sexism
- racism
- homophobia
- ableism
- colonial narratives
- class inequalities
- stereotypes and misinformation
AI absorbs these patterns and often reproduces them, sometimes subtly, sometimes blatantly. For social workers, this matters deeply. You might be working with young people, migrants, survivors of harassment, LGBTQIA+ communities, people with disabilities, or others who already face discrimination. A biased AI response can reinforce harmful narratives you are actively trying to dismantle. This isn’t a glitch. It’s baked into how the technology works.
AI lacks empathy, context and lived experience
Social work relies on:
- understanding relationships
- interpreting emotion
- recognising power imbalances
- being trauma-sensitive
- holding space for nuance
AI cannot do any of this. It doesn’t “understand” people, it predicts patterns in text. So even when its output looks polished, it may:
- miss the emotional weight of a situation
- misinterpret the meaning behind someone’s words
- offer advice that is tone-deaf or unsafe
- flatten complex lived experiences into generic summaries
In social work, where human connection is central, this can create real harm.
AI is confident, even when it’s wrong
One of the most dangerous things about AI is that it often sounds confident and authoritative even when it’s inventing information, known as “hallucinating.”
For example, it may:
- cite statistics that don’t exist
- describe laws inaccurately
- misinterpret trauma dynamics
- generalise based on stereotypes
If you use AI to prepare materials or understand an unfamiliar topic, it’s essential to verify the information elsewhere.
Privacy and dignity are non-negotiable
This cannot be said strongly enough:
No sensitive or personal information should ever be placed into an AI system.
That includes:
- personal stories
- identity-based details
- HR information
- anything that was shared in confidence
Most AI tools send data to external servers, where it may be stored, used to retrain models, or accessed by third parties. Even if a tool claims otherwise, the risk is simply too high for social work. People trust us with their stories. Protecting that trust must come before any convenience.
AI can weaken critical thinking if we’re not careful
AI often gives quick, neat answers, which can feel helpful. But social work rarely fits into neat answers. If we rely too heavily on AI-generated ideas or framing, we may:
- lose nuance
- accept oversimplified explanations
- miss systemic causes
- overlook marginalised perspectives
- rely on patterns instead of reflection
Social work needs slow thinking, curiosity, and complexity, all the things AI cannot provide on its own.
AI has social and environmental costs
It’s also important to remember that AI is not “digital magic.” Training large models uses enormous amounts of energy and water, and often depends on underpaid labour for content moderation. These invisible impacts matter, especially when working for social and environmental justice. Using AI thoughtfully means recognising that our digital choices have real-world consequences.
So… should social workers use AI at all?
AI can certainly play a role, but only when used critically, safely and with human judgment leading the way.
It can help with:
- drafting non-sensitive text
- summarising public documents
- simplifying policy language
- translating generic material
- brainstorming ideas you later refine
But it should never replace:
- empathy
- critical thinking
- relationship-building
- ethical judgment
- trauma sensitivity
- or the lived experiences of the people we work with
A final reflection
Being critical of AI isn’t about resisting technology or refusing to change. It’s about recognising that social work is, at its heart, human work. Our relationships, our listening, our intuition, our sensitivity to power and emotion, these are things no machine can replicate. Approaching AI with care protects the dignity of the people we serve and strengthens the integrity of our practice.
In our next blog, we’ll explore practical tips and gentle strategies for using AI responsibly in social work inspired by our own AI Policy at Fairspace.
References
- ILDA. “AI and Confusion: Why Intersectional Fairness Matters in AI, a Conversation with Ilina Georgieva from Diversifair.”
- National Skills Council (Malta). “Myths & Misconceptions About AI: What You Should Know.”
- OpenAI. “Why Language Models Hallucinate.” 2023.
- MIT Sloan School of Management. “When AI Gets It Wrong: Addressing AI Hallucinations and Bias.”
- Ontario College of Social Workers and Social Service Workers. “The Use of Artificial Intelligence in Practice.” Spring 2024.
- Loga, Raquel. “AI’s Cognitive Implications: The Decline of Our Thinking Skills?” IE Center for Health & Well-Being, 2025.
- Zewe, Adam. “Explained: Generative AI’s Environmental Impact.” MIT News, 2025.
- Bashir, Noman et al. “The Climate and Sustainability Implications of Generative AI.” MIT Climate Project, 2024.
- Magruder, Karen. “Ethics and Best Practices: Artificial Intelligence and Social Work Education.” Social Work Today, 2023.
- Hiltz, Barbara. “Editorial: Embracing AI in Social Work.” International Journal of Social Work Values and Ethics, 2025.
AI Assistance Acknowledgement: This blog was partly drafted with AI assistance for early drafting and structure. All content was reviewed and refined by a human author.
