Time to Move Beyond the ASWB Licensing Exams: Can Generative Artificial Intelligence Offer a Way Forward for Social Work?

Sep 21, 2026

Uploaded by William

Social work scholars have long questioned the validity and utility of the Association of Social Work Boards (ASWB) licensing exams. Data released in 2022 revealed severe disparities in pass rates based on race, age, and language, intensifying concerns. This paper tests generative artificial intelligence such as ChatGPT on ASWB practice questions to examine whether it reflects real-world social work reasoning.

LanguageEnglish

CategoryExam

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Abstract

Generative AI Models

ASWB Licensing Exams

Invited Article Time to Move Beyond the ASWB Licensing Exams: Can Generative Artiﬁcial Intelligence Offer a Way Forward for Social Work? Bryan G. Victor 1 , Sheryl Kubiak 1, Beth Angell2, and Brian E. Perron2 Research on Social Work Practice 2023, Vol. 33(5) 511–517 © The Author(s) 2023 Article reuse guidelines: [sagepub.com/journals-permissions](sagepub.com/journals-permissions) [DOI: 10.1177/10497315231166125](DOI: 10.1177/10497315231166125)[ ](DOI: 10.1177/10497315231166125)[journals.sagepub.com/home/rsw](journals.sagepub.com/home/rsw) Abstract Social work scholars have long questioned the validity and utility of the Association of Social Work Boards (ASWB) licensing exams. Data released in 2022 revealed severe disparities in pass rates based on race, age, and language, exacerbating these concerns. In this paper, we explore the potential of generative artiﬁcial intelligence (AI) such as ChatGPT to address core problems of the ASWB exams, including the use of a multiple-choice format that does not reﬂect real-world social work practice. To assess its social work reasoning, we used ChatGPT to answer ASWB-developed practice questions for the Bachelors, Masters, and Clinical exams. ChatGPT scored 76%, 80%, and 64%, respectively, and identiﬁed additional validity challenges. Based on this performance, we provide a proof-of-concept for how generative AI might move us toward a more valid and equitable exam. While we strongly support licensure requirements, state regulators and legislators should temporarily suspend the use of the ASWB exams for this purpose. Keywords licensing, social work professionals, generative artiﬁcial intelligence, validity, dynamic testing The emergence of generative artiﬁcial intelligence (AI) technologies and large language models (LLMs), such as ChatGPT, the new Bing, and Bard, are reshaping how we interact with information and communicate with others. Previous AI technologies have received some attention in social work, primarily in research settings for information extraction (e.g., Frey et al., 2020; Perron et al., 2019; Victor et al., 2021) and simulation-based clinical and job training (e.g., Asakura et al., 2020; Smith et al., 2023) . Generative AI represents a substantial leap forward and opens new and far-reaching opportunities for social work research, education, and practice. Recent headlines of ChatGPT passing high-stakes medical, business, and legal exams (Hammer, 2023; Kelly, 2023) inspired our decision to evaluate its performance on the social work licensing exams. We think that generative AI can offer additional insights into recent discussions of whether the profession’s reliance on licensing exams developed by the Association of Social Work Boards (ASWB) is appropriate, effective, and ethical for their intended purpose (Apgar & Nienow, 2023; DeCarlo, 2022; NASW, 2023; Nienow et al., 2023) . To set the stage for this discussion, we ﬁrst provide a plain language overview of generative AI and the ASWB exams. We then detail our evaluation of ChatGPT’s performance on ASWB-developed practice exam questions. We discuss critical validity issues that our evaluation helped reveal. Combined with prior research, we believe our ﬁndings indicate that the problems of the ASWB exams do not justify continued use, and the emergence of generative AI models offers a potential solution for creating innovative and equitable procedures for evaluating the preparedness of social workers to practice ethically and safely. Generative AI Models Whereas traditional AI technologies rely on rules-based programming to understand and respond to user input, generative AI models use a combination of deep learning and natural language processing to understand the nuances of human conversation and respond with natural, human-like language. LLMs are a speciﬁc type of generative AI trained on massive datasets of text, such as books, articles, and web pages. These models use deep learning techni

FAQs

Why do the authors question the ASWB licensing exams?

They note long-standing concerns about the exams’ validity and utility, amplified by 2022 data showing disparities in pass rates across race, age, and language.

How did the authors evaluate ChatGPT’s performance?

They used ChatGPT to answer ASWB-developed practice questions for the Bachelors, Masters, and Clinical exams, then compared scores and identified additional validity challenges.

What scores did ChatGPT achieve on the ASWB practice exams?

ChatGPT scored 76% for the Bachelors exam, 80% for the Masters exam, and 64% for the Clinical exam, while also revealing validity problems.

Sep 21, 2026

Uploaded by William