July 2026Vol. 1, No. 2AI in Child Welfare: A Spotlight on Workforce Development and Analytics

As AI becomes embedded in everyday life, it is increasingly part of the conversation in child welfare. AI tools and AI-enabled processes have the potential to help child welfare agencies reduce administrative burden and make better use of existing data to improve support for children and families. The Administration for Children and Families (ACF) provides guidance for using AI as a tool that can improve operations and support decision-making when implemented thoughtfully and responsibly.

There are many ways AI can help child welfare professionals streamline routine tasks and administrative workloads, potentially allowing staff more time and energy to focus on supporting families. For example, agencies may explore using AI to help draft materials, develop training resources, summarize complex information, or synthesize data across systems. At the same time, it is critical to recognize the risks and limitations associated with AI use in child welfare settings, such as concerns related to data privacy and confidentiality, data quality, staff understanding of AI tools, and overreliance on automated outputs. ACF emphasizes the importance of human oversight to keep a "human in the loop" so that AI is used as a tool for decision support and not a replacement for professional judgment. This is particularly vital in high-stakes environments affecting children and families.

Agencies continue to explore ways to strategically apply AI, and an emerging area of interest is AI use in workforce development and analytics, including onboarding, training, supervision, quality monitoring, and retention. As the landscape of AI use evolves, there will continue to be a need for guidance that reflects ongoing research and the realities of child welfare practice.

Featured Federally Funded Project

The Children's Bureau (CB) is supporting ongoing guidance in this area through the Quality Improvement Center for Workforce Analytics (QIC-WA), a CB-funded grant recipient focused on strengthening workforce decision-making through data and analytics.

Much of the QIC-WA's work involves translating the complex subject of workforce analytics into practical, digestible tools for child welfare professionals and leaders. Using these resources, agency leaders and human resources professionals can better understand what data to collect, measure, and use to improve and support their workforce. The QIC-WA's Illuminating Pathways tool, for example, pairs short podcast-style audio segments with a video and frequently asked questions to help leaders recognize the educational and work experience paths that lead to child welfare careers, information that can then be used to strengthen recruitment, hiring, and retention efforts. This is essential, as a healthier child welfare workforce contributes to better outcomes for children and families. Another recent example is the QIC-WA's Workforce Analytics Capacity Assessment tool, a self-assessment resource that helps agencies examine six areas of workforce analytics capacity: culture, knowledge and skills, technology, data, accessibility and utility, and practice. Leaders can use the resulting ratings and recommended next steps to respond to workforce trends like turnover and staffing shortages and create an actionable roadmap for improvement. The QIC-WA's newest tool, the Workforce Analytics Assistant, is a web-based application that helps agencies identify hiring and retention trends by turning their workforce data into customized metrics, visuals, and AI-supported recommendations. Data entered into QIC-WA tools are never saved by the project. Explore more of the QIC-WA's resource offerings on its website.

The following sections of this article reflect insights from the QIC-WA. Read on to learn about the QIC-WA's research-based perspectives on AI-enabled workforce analytics and considerations for public child welfare agency workforce development, human resources, and program leaders regarding implementation aligned with Federal guidance and ethical practice.

Partner Spotlight: AI in Child Welfare Workforce Development and Analytics: Benefits, Risks, and Examples

Contributed by Robert D. Blagg, Ph.D.
Quality Improvement Center for Workforce Analytics, Co-Principal Investigator
Agile Visual Analytics Lab, Director

Why This Matters Now

For many years, workforce-relevant data in public child welfare agencies have often been spread across human resources (HR) systems, learning management systems, and Comprehensive Child Welfare Information Systems (CCWIS) environments, and long-standing data quality issues have limited the usefulness of workforce analytics for managers (Children's Bureau, 2018). This has made it difficult for agencies to use data to support key workforce strategies such as onboarding, training, and professional development—areas in which agencies have invested heavily. National workforce data indicate that agency leaders view supportive supervision and professional development as key retention strategies (Elgin et al., 2025), and leading organizations have leveraged data to execute these strategies. As child welfare agencies improve the quality and connectedness of their data systems, AI now offers the potential to help agencies unlock solutions to these long-standing problems, but only when it is implemented with clear governance, privacy protections, and human accountability (National Institute of Standards and Technology [NIST], 2024).

What AI-Enabled Workforce Analytics Means in Practice

AI-enabled workforce analytics refers to the application of machine learning, natural language processing, and generative AI tools to strengthen workforce development decision-making across the employee lifecycle. In public child welfare agencies, this can occur in many ways, including the following:

Extracting consistent, actionable insights from unstructured information. AI systems can analyze narrative supervision notes, coaching transcripts, recorded practice sessions, and case documentation to identify patterns in worker behavior, fidelity to practice models, and developmental needs. For example, machine learning tools have been used to automatically code motivational interviewing sessions and generate structured fidelity feedback that would otherwise require trained human raters (Imel et al., 2019). Similar AI-assisted fidelity monitoring approaches have been deployed in child welfare and prevention settings to assess service quality at scale, increasing consistency and reducing supervisory burden (Gourley, 2022a, 2022b). In workforce analytics terms, this allows agencies to move from anecdotal impressions of staff performance to systematic, comparable indicators derived from routine practice data.
Automating routine reporting, data integration, and dashboarding. Public child welfare agencies operate across fragmented systems (e.g., case management information systems, human resources information system [HRIS], or learning management system platforms), often requiring manual compilation of workforce reports. AI-enabled analytics tools can automate the extraction and synthesis of cross-system data to generate real-time dashboards on hiring pipelines, onboarding completion, amount and frequency of training, supervision frequency, caseload distribution, overtime patterns, and turnover risk indicators. The federal CCWIS framework explicitly emphasizes the importance of automated data quality checks and advanced analytics to improve agency decision-making (Children's Bureau, 2018).
Creating faster feedback loops for supervisors, trainers, and HR teams. Traditional workforce analytics in child welfare are often quarterly or annual (e.g., turnover reports, climate surveys). AI systems can shorten feedback cycles by continuously analyzing onboarding survey responses, pulse check-ins, training reflections, supervision logs, and exit interview narratives to surface emerging themes. Research in implementation science demonstrates that structured feedback systems and fidelity monitoring are associated with improved workforce stability and reduced turnover when implemented supportively (Aarons et al., 2009). AI systems can accelerate these feedback processes by flagging early warning indicators of burnout (e.g., increased after-hours documentation, sentiment shifts in reflective notes, declining engagement survey responses), thereby enabling timely, preventive supervisory interventions.

AI-enabled workforce analytics should not replace professional judgment. Rather, it can augment supervisory insight, increasing consistency in quality monitoring, reducing administrative burden, and accelerating learning cycles—provided that governance structures ensure transparency and supportive (rather than punitive) use.

Examples From Child Welfare Practice

Wyoming: District-level segmentation to target training. A Government Technology article describes Wyoming's Department of Family Services using AI for fidelity monitoring and expanding it so staff can segment large data sets by district, replacing a manual process (Edinger, 2025). The same article notes that confidentiality requirements shaped implementation (e.g., release forms before uploading recordings) and that the resulting analytics inform statewide motivational interviewing training priorities (Edinger, 2025).

District of Columbia: A CCWIS "help" chatbot for policy and procedure questions. The District of Columbia's Child and Family Services Agency (CFSA) published the CFSA AI Values Alignment Report, which describes a generative AI chatbot (Case Operations Resource Assistant, or CORA) intended to provide staff with immediate answers about policies, procedures, and system functionality. The report also states that CORA draws only from approved documentation and not confidential case files (CFSA, n.d.).

Benefits and Risks for Workforce Development and Analytics

Across these kinds of use cases, HR and workforce leaders typically see the following five benefit areas—each paired with a risk that should be managed:

Scaling coaching and fidelity monitoring across units and providers. Risk: workforce distrust if the system is perceived as monitoring or potentially punitive, rather than as supportive coaching (Aarons et al., 2009; Imel et al., 2019).
Turning training into actionable analytics (e.g., skill growth over time, unit-level patterns). Risk: false precision and overreliance on metrics that were not designed for performance management (NIST, 2024).
Reducing administrative burden by improving access to policies and "how-to" guidance during the course of work. Risk: inaccurate or outdated guidance if knowledge sources are not governed and refreshed (CFSA, n.d.).
Integrating data across HR, training, and CCWIS to support decision-making. Risk: AI can miss or amplify existing data quality problems; CCWIS guidance emphasizes agency responsibility for data quality planning and review (Children's Bureau, 2018).
Using generative AI for summarization or synthesis to shorten time needed to draft reports. Risk: "confabulation" or "hallucination" (confident but incorrect outputs) and other generative AI-specific risks; agencies should require review and maintain auditability (NIST, 2024).

Getting Started With AI-Enabled Workforce Development and Analytics

The following presents a compact, practical set of first steps for agencies that are exploring early implementation (NIST, 2024):

Scope: Start with a narrow workforce development use case (e.g., training feedback, onboarding support) and explicitly prohibit use for automated or immediately consequential decisions (e.g., discipline) (Office of Management and Budget, 2025).
Prepare: Document data sources, consent and authority to use, data governance, and data quality checks before building dashboards or implementing AI models (Children's Bureau, 2018).
Operationalize success: Define what "good" looks like (accuracy, workload reduction, efficiency, or cost savings) and test in a small pilot before scaling (NIST, 2024).
Establish guardrails: Create a people-driven oversight process: human review of outputs, logging, and a procedure for staff to challenge or correct errors (NIST, 2024).
Ensure sustainability: Plan change management: communicate purpose, limits, and how staff feedback will shape development and implementation (Children's Bureau, 2018).

As agencies explore implementation of AI‑enabled workforce processes, effective and sustained change management depends on building staff buy‑in and support by engaging a representative group of staff with different roles and perspectives.

Key Takeaways

Child welfare agencies are already beginning to use AI to scale evidence-based practice coaching and fidelity monitoring, support child welfare practice quality assurance, and provide staff with policy and procedure guidance within CCWIS environments (CFSA, n.d.; Edinger, 2025; Gourley, 2022a, 2022b). The most promising near-term workforce analytics applications of AI in public child welfare agencies are those that are narrowly scoped, transparent in design, and auditable in operation, such as automated summarization of supervision notes into structured coaching themes; natural language processing of onboarding surveys to identify common training gaps; and real-time dashboards integrating HRIS, CCWIS, and learning management data to monitor caseload distribution, overtime, and early tenure attrition (Imel et al., 2019). These applications improve supervisory insight and accelerate feedback loops without removing the humanity from consequential employment decisions. Consistent with federal AI risk management guidance, such tools should include documentation of data sources, model limitations, and human oversight checkpoints to ensure they strengthen workforce development rather than replace professional judgment or erode workforce trust (NIST, 2024).

References

Aarons, G. A., Sommerfeld, D. H., Hecht, D. B., Silovsky, J. F., & Chaffin, M. J. (2009). The impact of evidence-based practice implementation and fidelity monitoring on staff turnover: Evidence for a protective effect. Journal of Consulting and Clinical Psychology, 77(2), 270–280. https://doi.org/10.1037/a0013223

Children's Bureau. (2018, November 28). Comprehensive Child Welfare Information System (CCWIS) technical bulletin #6: CCWIS data quality plan. U.S. Department of Health and Human Services, Administration for Children and Families. https://acf.gov/cb/training-technical-assistance/ccwis-technical-bulletin-6

District of Columbia Child and Family Services Agency. (n.d.). CFSA AI values alignment report. https://techplan.dc.gov/sites/default/files/dc/sites/itstrategicplan/publication/attachments/OCTO%20Submission%20AI%20Values%20Alignment%20Report%20-%20CORA%20Use%20Case%20FINAL%20FOR%20PUBLICATION.pdf

Edinger, J. (2025, March 14). Govt. considerations for adding AI into human services. Government Technology. https://www.govtech.com/artificial-intelligence/govt-considerations-for-adding-ai-into-human-services

Elgin, D. J., Barbee, A. P., McCarthy, M. L., Kluckman, M., Ringeisen, H., & Dolan, M. (2025). Child welfare workforce onboarding, training, and professional development from 2021 to 2022 (OPRE Report #2025-078). U.S. Department of Health and Human Services, Administration for Children and Families, Office of Planning, Research, and Evaluation. https://acf.gov/sites/default/files/documents/opre/opre-child-welfare-workforce-onboarding-aug25.pdf

Gourley, C. (2022a, May 10). Lyssn AI to help assess the quality of prevention services offered under Family First Act. Business Wire. https://www.businesswire.com/news/home/20220509006294/en/Lyssn-AI-to-Help-Assess-the-Quality-of-Prevention-Services-Offered-Under-Family-First-Act

Gourley, C. (2022b, November 10). DC Child and Family Services Agency to use artificial intelligence in support of child welfare programs. Business Wire. https://www.businesswire.com/news/home/20221110005177/en/DC-Child-and-Family-Services-Agency-to-Use-Artificial-Intelligence-in-Support-of-Child-Welfare-Programs

Imel, Z. E., Pace, B. T., Soma, C. S., Tanana, M., Hirsch, T., Gibson, J., Georgiou, P., Narayanan, S., & Atkins, D. C. (2019). Design feasibility of an automated, machine-learning based feedback system for motivational interviewing. Psychotherapy, 56(2), 318–328. https://doi.org/10.1037/pst0000221

National Institute of Standards and Technology. (2024). Artificial intelligence risk management framework: Generative artificial intelligence profile (NIST AI 600-1). U.S. Department of Commerce. https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf

Office of Management and Budget. (2025, April 3). Accelerating federal use of AI through innovation, governance, and public trust (Memorandum M-25-21). https://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of-AI-through-Innovation-Governance-and-Public-Trust.pdf