When professional associations publish AI competency frameworks, the temptation is to skim the document and move on. The new ACRL AI Competencies for Academic Library Workers, approved by the ACRL Board in October 2025, is worth reading carefully. Not just by librarians. By anyone in higher education who’s been waiting for a thoughtful, profession-grounded answer to the question of what AI literacy should mean for working professionals.
AI Competencies for Librarians
The task force, co-chaired by Keven Jeffery and Jason Coleman, organizes the document around five guiding mindsets and four competency categories. The mindsets come first, before any concrete skills: curiosity, skepticism, judgment, responsibility, collaboration. The four competency categories follow the structure Ng et al. (2021) developed in their content analysis of AI literacy literature: Ethical Considerations, Knowledge & Understanding, Analysis & Evaluation, Use & Application.
Most AI competency frameworks I’ve covered, like UNESCO‘s (2024) framework for students, treat AI literacy as primarily about adopting AI well. ACRL goes further. The task force makes a refusal-preserving move that should be more common in professional documents. AI literacy doesn’t mean adopting AI uniformly. Library workers keep the right to opt out of specific AI tools when ethics, performance, or institutional values don’t line up.
The framework states this directly: “adoption of AI technologies is neither necessary nor beneficial in all cases” (p. 8). That’s a strong stance for a competency document to take. It’s the difference between a framework that prepares workers to work with AI and one that prepares workers to choose whether to work with AI.
The same logic should apply to teachers. A K-12 or higher ed AI competency framework that doesn’t include the right to refuse leaves educators feeling like the only acceptable response to AI is enthusiastic adoption. That’s a bad starting point for genuine pedagogical thinking.
AI Detection Gets a Public Verdict
The task force notes that “tools claiming to detect AI-generated writing are not completely accurate and can be circumvented” (p. 6). That sentence appears inside a baseline competency. Library workers are expected to understand it.
The empirical case for that statement is strong. Russell, Karpinska, and Iyyer’s (2025) work on human and automatic AI detection, which I’ve covered before, showed that frequent LLM users beat almost every commercial detector, while open-source detectors collapsed under paraphrasing and humanization. ACRL is now writing that finding into a professional competency framework. Institutions buying AI detection software are doing so against the advice of one of the most relevant professional bodies in higher education.
What Teachers Should Borrow
The framework is structured for librarians, but four moves transfer cleanly to classroom teaching contexts.
The first move worth borrowing is the mindset-before-skill ordering. Teachers thinking about AI integration should start with dispositions, not tool training. The five mindsets are exactly what students need to develop too.
Then there’s the ethics-first framing. ACRL’s choice to lead with Ethical Considerations is a useful template for educators. Teaching AI ethics should never be the last thing the curriculum covers.
The technical literacy bar surprised me. It’s higher than I expected. Library workers are asked to understand retrieval-augmented generation, agentic AI, and the difference between generative, discriminative, and predictive AI. Teachers can’t assess what students are doing with these tools without similar literacy. Chiu’s (2025) editorial on the AI literacy versus competency split makes the same point in different language.
Finally, the values-laden endorsement of open source matters. The framework points out that “open-source AI models can align with library values by promoting transparency, community-driven innovation, and broad access to technology” (p. 4). Education has its own version of this argument. Tools that enable transparency, accessibility, and freedom from vendor lock-in often beat equivalent proprietary products.
Where the Framework Falls Short
The document is honest about its expiration date. AI moves fast. The framework avoids product names and specific job functions to extend its useful life, but the task force says no competency framework will stay current for decades, or even a few years.
I’d add one concern. The framework is strong on individual competencies but light on institutional accountability. Library workers are expected to navigate AI systems that institutions integrate without consulting them. Naming that problem is important. The framework doesn’t say much about what library workers should do when their institutions integrate AI badly. That work has to happen elsewhere.
References
- Association of College and Research Libraries. (2025). AI competencies for academic library workers. American Library Association. https://www.ala.org/sites/default/files/2025-10/acrl_ai_competencies.pdf
- Chiu, T. K. F. (2025). AI literacy and competency: definitions, frameworks, development and future research directions. Interactive Learning Environments, 33(5), 3225–3229. https://www.tandfonline.com/doi/full/10.1080/10494820.2025.2514372
- Russell, J., Karpinska, M., & Iyyer, M. (2025). People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text (arXiv:2501.15654). arXiv. https://arxiv.org/abs/2501.15654
- UNESCO. (2024). AI competency framework for students. United Nations Educational, Scientific and Cultural Organization. https://doi.org/10.54675/JKJB9835