AI Analyzes the Social Work Licensing Exam, Concerns Deepen

By The Imprint.

Summary: Investigative report on concerns raised after AI analysis of the social work licensing exam.

Relevance: Speaks directly to social work licensure and, given the racial pass-rate gap the researchers cite, to the Inequality & Algorithmic Harm subdomain: uses AI to interrogate the validity of the ASWB exam that gatekeeps entry into the profession.

Key findings: Michigan State researchers found ChatGPT often gave “a compelling and sometimes better rationale” than the ASWB’s official answer key, faulting the exam for outdated concepts and a multiple-choice format that can’t capture nuanced ethical practice. They cite ASWB’s own data showing 45% of Black test-takers passing on the first attempt versus 84% of white test-takers, and call for states to “temporarily suspend the exam” pending redesign; ASWB disputes this, maintaining its questions are “thoroughly reviewed.”

Read the original: https://imprintnews.org/top-stories/ai-analyzes-the-social-work-licensing-exam-and-concerns-deepen/241949