AI Resilience Report forSocial Workers, All Other

Social work tasks are very hands-on and personal, so there are no widespread AI tools replacing social workers. Most of their work – like talking with families, giving advice, and coordinating support – requires empathy and judgment. In fact, news reports show that attempts to use algorithms in child welfare have hit problems.

For example, an AP News article described a case where an AI screening tool labeled some disabled parents as “high risk” simply because of their disability ^[1]. This kind of mistake shows that AI still struggles with the human side of social work. Career resources like O*NET even note that for “Social Workers, All Other,” there isn’t a fixed list of tasks to automate because the work varies so much ^[2].

In short, aside from basic data systems or scheduling apps, there isn’t any popular AI that can do a social worker’s core job for them, so most day-to-day activities remain human-led.

Because social work is so people-focused, AI is being adopted slowly and carefully in this field. There aren’t many off-the-shelf AI products for caseworkers, and social service budgets are often tight. Projects that have tried AI see mixed results.

One example: an AI “matchmaking” tool for foster care ran in several states and ended up improving almost no outcomes ^[1]. When new tech didn’t help much, agencies stayed cautious. There are also strong ethical and legal concerns.

For instance, a county child-welfare screening tool in Pennsylvania drew a Justice Department review over bias ^[1]. People expect fairness and personal trust in social services, so any hint of error or unfairness makes organizations nervous about using AI.