In the final stage of CAM, students are encouraged to explore and apply skills independently in new contexts (Collins et al., [ 7 ]). The role of the teacher fades, and the students can take on independent projects with all the acquired knowledge and skills they have gained through the stages of the cognitive apprenticeship with their teacher.

During the reflection phase of the CAM, students focus on reflecting on their performance and comparing it to examples from the experts (Collins et al., [ 7 ]). Reflection emphasizes the importance of students pausing and reflecting on their learning experiences, examining their thought processes, and assessing their approach to a given task or assignment. Research studies show that reflective practices can further enhance students' understanding through the self-evaluative processes and contribute insights to future performance (Ma et al., [15]). In the context of AI usage for writing, reflection can help, especially for those students who may lack awareness of how AI interactions influence their processes and writing outcomes (Oh et al., [17]). Furthermore, reflective practices encourage students to think critically about the ethical dimension of AI usage.

students verbalize their understanding to solidify their learning (Collins et al., [ 7 ]). The teacher's role shifts from direct instruction or guidance to encouraging students to express their thinking (or making it visible), often through questions that can take their understanding to a deeper level. At the articulation phase, focus on students expressing their understanding of how, when, and why to use AI.

In the scaffolding stage of CAM, teachers gradually remove support as students gain competence in a new skill (Collins et al., [ 7 ]). Scaffolding as a strategy fosters confidence in students and encourages independence to work beyond their current ability and beyond the parameters of the assignment (van de Pol et al., [21]). Scaffolding can enhance students' writing abilities and composition structure, contributing to better writing outcomes and enhancing their confidence as writers (Sidky, [19]). Scaffolding in AI-supported writing instruction can effectively structure learning and content development (Hui & Sprouse, [11]).

In the coaching stage of CAM, teachers provide guidance and feedback as students perform the task (Collins et al., [ 7 ]). Research studies focused specifically on AI literacy and use among students highlight that hands-on learning strategies supported by teacher feedback and coaching can enhance problem-solving abilities and better prepare students for future interactions with AI (Escalante et al., [ 9 ]; Sinha et al., [20]). Coaching and feedback are also essential in helping students navigate challenges, especially around ethical consideration and authenticity (Cardon et al., [ 4 ]).

The initial stage of CAM focuses on teachers demonstrating clear steps for the task or skill that students are to acquire (Collins et al., [ 7 ]). Several studies highlight that modeling AI use with an emphasis on making connections with students' everyday experiences can enhance their understanding of the AI applications (e.g., Dai et al., [ 8 ]).

ethical uses

articulation stage

prompt engineering

scaffolding

Cognitive Apprenticeship Model (CAM)

This chapter has advocated for using a scaffold such as the Cognitive Apprenticeship Model (CAM) as a theoretically grounded framework to support educators in navigating this shifting landscape. CAM emphasizes the explicit modeling, coaching, scaffolding, articulation, reflection, and exploration strategies needed in the demands of teaching scholarship and writing in an AI-supported learning environment.

It is worth noting briefly that though much of the attention here has been to ChatGPT in particular, and that that particular instance was chosen via its global popularity and accessibility, that many other AI LLM instances exist, with the number growing every day.

ChatGPT can enhance the feedback loop that students experience in their writing program. For example, ChatGPT gives instant, context-sensitive feedback on all aspects of writing. It will tell students how they can improve sentence structure, coherence, or argumentative strength. The immediacy of the feedback is a legitimate strength as it allows students to make revisions in real time in a way that reinforces learning and allows them to make corrections and improvements as they write. ChatGPT can help students identify mistakes, but it can also give students guidance on how to improve specific aspects of a draft. For example, if a student has written a weak conclusion, ChatGPT might suggest specific strategies to restate the thesis and reinforce the argument effectively.

"teach students the key features of effective writing so they can use them in their own writing" (p. 24). In this scenario, argumentative essays will be used as the lesson. The teacher can provide a link for students, such as The Daring English Teacher's ([34]) "101 Argument Essay Prompts for High School." Students can pick a topic, then prompt the LLM with the topic and the criteria.

One of the examples of integrating writing and reading that What Works Clearinghouse ([43]) gives is with the objectives of students recognizing cause and effect structure when it appears in their reading and applying a cause and effect structure to their own writing (p. 33). Students can use an LLM, such as ChatGPT or a teacher-customized chatbot in MagicSchool AI, to analyze something they have read.

The time saved via the above interventions provide significantly more time for students to reflect orally and in writing, individually and collectively about their writing, product and process, but AI/LLM can also be involved more directly, with students querying the AI/LLM as to the quality/fidelity of co-created work according to a rubric or set of instructions, either on their own work, on that of another student, or both. Students could then reflect on their own assessment of the work, product and process, and the degree to which AI/LLM improved, interfered with, or was neutral to, their normal writing processes, and how they might go about such implementation when and if they are allowed more choice on the degree to that tech involvement in the future.

student inspiration doesn't spark in these settings, is slow to move from ideation to actual drafting, or produces very little work so that teachers can't intervene in ways which are likely to create high-level differentiation between lab-based work and work students complete entirely at home.

The real-time speed of composition of AI/LLM such as ChatGPT means that teachers not only can produce "student" models in advance of class, but they can also do so in class, often in seconds, providing student-like writing available for critique and edit.

The third recommendation is to "use assessments and student writing to inform instruction and feedback"

The second recommendation is to "integrate writing and reading to emphasize key writing features"

The first recommendation is to "explicitly teach appropriate writing strategies using a Model-Practice-Reflect instructional cycle"

The final stage of writing talent development is the integration stage where the knowledge and skills acquired in the precision stage a synthesized into a holistic understanding and the emergence of personal voice and style. In this stage, the students integrate the parts into a whole as they master their craft and hone their artistry. They transition from someone who has writing skills to the deeper identity as a writer.

The initial stage of writing talent development is the romance period where an individual develops curiosity and excitement about a topic or a performance space. The romance stage is fundamentally an initial motivation phase where attention is directed toward learning and performing in the target domain, in this case creative writing. During the romance stage, the student is driven by wonder, imagination, and a sense of exploration.

The second stage of writing talent development is the precision stage where learners practice and master the details, conventions, and technical aspects of writing for a literary purpose. This stage of development is characterized by rigorous study and practice. Students develop the discipline of writing routines, and the craft of conventional and innovative language structures.

AI/LLM co-creation in schools and associated concerns and opportunities

co-creation

concerns about proper teacher training and support, subscription costs for some AI/LLM models, student hyper-dependence, ongoing, often significant, inaccuracies and errors created by AI/LLM, and other developmental and implementation issues.

Reflecting on writing with an AI/LLM

Practicing writing with an AI/LLM

Modeling writing with an AI/LLM

Co-creation is only now starting to enter this literature base

LLM

Our proposed adapted model of creative writing intervention highlights the concept of co-creation in the writing domain, which is especially beneficial for middle- and high-school students in the areas of inspiration, creativity, and engagement, countering the unfortunately common belief that the use of AI tends to reduce student creativity.