MANAGEMENT UPDATE.

A GUIDE TO AI AND CHILD WELFARE

With the need to support increasing caseloads and navigate changing policy frameworks, the child welfare workforce faces an increasingly complex task of making key decisions for children and their families, according to a new report from the IBM Center for the Business of Government.

State and Local Management requirements, along with federal rules, means documenting actions; complying with mandates from different layers of government and delivering high-stake performance outcomes.

The report by David Schwartz, an AI Policy Practitioner and Social Work Researcher, points to the “need for workers who can read a dangerous situation in seconds, and for workers who can maintain meticulous records.”

As he writes, “the gap between what the child welfare workforce has the responsibility to do, relative to what existing systems enable them to do, has widened in recent years. To close that gap, states are beginning to seek help through responsible application of AI.

Based on a roundtable discussion, convened by the IBM Center and the Michigan School of Social Work, key use cases emerged that illustrate the potential of AI to help deal with the difficult issues that confront the world of child welfare.

Here they are, with the easiest to tackle right now, coming first and the more aspirational ones listed later.

“Helping new workers understand policies and procedures. If a worker needs to find the right form or understand their agency’s policies, AI can help with that. This seemed universally accepted by the group.”

“Helping workers sift through enormous case files to find what they need, using tools for smart search through documentation. When a child has been in the system for years, the paper trail is massive. AI can help workers find the needle without reading the entire haystack.”

“Dealing with “tasks like redacting sensitive information from files, work that currently requires human hours but not human judgment.”

“How to prepare these new workers before they are out in the field making real decisions about real families? AI-powered simulations and training scenarios came up repeatedly as promising territory.”

“Using AI to identify the children at highest risk of serious injury or death. Finding those needles in the haystack. No one wanted to abandon this goal, which is too important. But there was clear consensus that government is not ready yet.”

One significant challenge pointed out in the report is that there is no standing forum to address ethical AI and child welfare specifically. As a result, “Agencies operate in silos because there’s no infrastructure for continuous learning and collaboration. More importantly, when agencies do come together to discuss AI, it’s often at conferences with many sales presentations—companies offering solutions, rather than practitioners sharing effective practices and lessons learned.”

Clearly there are hazards in the use of AI, ranging from bias to the concern of losing jobs. The discussion addressed these and other concerns and the report listed ten principles that agencies at any stage in their progress toward ethical and productive use of AI. They recommended that child welfare agencies: should: