Prompt engineering has changed considerably since Leo Lo introduced the CLEAR framework in 2023. Current AI models can interpret less structured requests, maintain longer conversations, and ask users for clarification. Even so, the framework still offers teachers and students a useful way to think about what makes an effective prompt.
CLEAR stands for concise, logical, explicit, adaptive, and reflective. These five principles encourage users to clarify what they want, organize their instructions, revise weak prompts, and evaluate the resulting output. In educational settings, that process matters because a well-written prompt can make the learner’s reasoning visible.
Lo developed the framework primarily for academic librarians and people beginning to work with generative AI. It is a conceptual framework, so it should be treated as a practical starting point instead of an empirically validated model of learning.
The Five Letters
Lo’s CLEAR Framework is a mnemonic for prompt engineering: Concise, Logical, Explicit, Adaptive, Reflective. Each letter names a core principle. Concise calls for stripping prompts of superfluous wording. Logical asks for ideas to follow a natural progression. Explicit specifies the output format and scope. Adaptive treats prompts as drafts that get revised when results miss the target. Reflective closes the loop, with users evaluating outputs and feeding what they learn into future prompts.
The Concise principle gets the cleanest example contrast in the paper. Lo argues that “a concise prompt removes superfluous information, allowing AI language models to focus on the most important aspects of the task, resulting in more pertinent and precise responses” (p. 2). The before-and-after example is “Can you provide me with a detailed explanation of the process of photosynthesis and its significance?” becoming “Explain the process of photosynthesis and its significance.” Less wording, sharper output.
Prompt Engineering CLEAR framework
What I find most useful in CLEAR is the Reflective component. Most prompt engineering advice treats prompting as a one-shot transaction. Lo (2023) emphasizes that “adopting a reflective perspective enables users to evaluate the performance of their AI model based on user feedback and their own assessments, identifying areas for improvement and adjusting their approach accordingly” (p. 2). That’s the meta-skill the framework builds toward: prompt engineering as iterative practice.
The Adaptive principle is also stronger than it looks. The willingness to revise a prompt when it doesn’t land is something many students still don’t do. They write one prompt, see a mediocre output, and accept it. The framework names that as a literacy gap, not a tool problem.
Cheng et al.’s (2025) work on AI questions and writing quality, which I’ve covered before, makes a parallel point: students who ask better questions of generative AI produce better writing. CLEAR gives teachers structured language for what better questions look like.
What Three Years Has Changed
The paper is upfront about its moment. April 2023 was very early. ChatGPT was at version 3.5 for most free-tier users. GPT-4 had just launched the month before. Reasoning models, agentic AI, retrieval-augmented generation, and the modern multimodal tools didn’t exist as mainstream products yet.
Three years on, the technical context is unrecognizable. AI tools handle clarification, ambiguous prompts, and follow-up questions much more elegantly. CLEAR’s emphasis on conciseness now reads as a workaround for limitations the models no longer have.
The other shift is conceptual. Prompt engineering itself is now a smaller piece of AI literacy than it was in 2023. Students need to think about hallucination, bias, source attribution, and ethical use in ways CLEAR doesn’t really address. The 2025 ACRL competencies framework Lo chaired, which I’ve covered before, is built around using AI thoughtfully across an entire information ecology. That’s a much wider lens.
What Teachers Should Take
Even with the temporal gap, three things in CLEAR hold up for classroom teaching.
The mnemonic structure works. Five letters that students can recall under pressure beats a fourteen-page framework. Teachers can use CLEAR as a fast scaffold for prompt engineering segments in K-12 or higher ed AI literacy modules.
The reflection step deserves emphasis in any teaching context. Most students don’t iterate on prompts because they haven’t been taught that iteration is part of the skill. Even two rounds of prompting with explicit reasons for revision is a high-leverage move.
The adaptive principle pairs well with explicit instruction on AI temperature settings, where they’re available. Students need to learn when to ask for tightly focused output and when to invite creative exploration. That’s a calibration capacity worth developing deliberately.
Limitations of the CLEAR Framework
CLEAR is memorable and accessible, but it does not address every part of responsible AI use. A prompt can satisfy all five principles and still produce an inaccurate, biased, or fabricated response. Students therefore need verification strategies alongside prompting strategies.
The framework also offers limited guidance on attribution, privacy, copyright, disciplinary differences, and the ethical boundaries of AI-assisted work. These issues have become increasingly important as generative AI has moved into routine educational practice.
Another limitation is the framework’s conceptual status. Lo presented CLEAR as a practical model for prompt construction, without testing whether its use improves learning, critical thinking, or long-term knowledge retention. Future research could compare CLEAR-based instruction with other approaches and examine which elements produce meaningful educational benefits.
I would therefore use CLEAR as an introductory scaffold. It gives students a manageable vocabulary for constructing and revising prompts. Once they understand that process, instruction should expand toward source verification, critical evaluation, transparency, and informed decisions about when AI use supports the learning goal.
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
- Lo, L. S. (2023). The CLEAR path: A framework for enhancing information literacy through prompt engineering. The Journal of Academic Librarianship, 49, Article 102720. https://doi.org/10.1016/j.acalib.2023.102720
- Lo, L. S. (2024). Evaluating AI literacy in academic libraries: A survey study with a focus on U.S. employees. College & Research Libraries, 85(5). https://crl.acrl.org/index.php/crl/article/view/26409/34344
