The rapid integration of Artificial Intelligence (AI) in education has created opportunities to reimagine

via platforms such as ChatGPT, Gamma, and Canva. These features help reduce cognitive load, generate ideas,

and sustain writing momentum. Its relevance to language education lies in addressing common challenges in

academic writing, such as organisation, linguistic confidence, and writing-related anxiety, while fostering long-

term development of academic literacy skills. The study was conducted with both tertiary and secondary ESL

students, capturing perspectives across different stages of language learning. A Google Form survey revealed

that learners reported significant benefits, including reduced writing anxiety, enhanced efficiency and increased

confidence in producing written work. At the same time, concerns were raised regarding over-reliance and

ethical use, highlighting the importance of responsible integration and clear guidelines. The study concludes

that ScaffoldED AI can serve as a pedagogical ally in secondary and higher education, empowering learners

with greater autonomy, reducing barriers to effective writing, and enriching teaching practices

thinking. ESL learners in particular struggle with writing anxiety, cognitive overload, and lack of personal

voice, which hinder performance and motivation.

This project introduces ScaffoldED AI, an AI-based approach designed to address these challenges while

aligning with established learning theories. Grounded in process-based instruction (Hyland, 2019), the

ScaffoldED AI rethinks AI not as a replacement for human thought, but as a thinking partner and scaffold for

improving clarity, reasoning, and autonomy.

Educators face pressing challenges in integrating AI into pedagogy while safeguarding learner agency and

creativity. Without appropriate scaffolding, the use of AI in writing classrooms can lead to several issues. First,

These concerns highlight the urgent need for innovations that embed structured scaffolding into AI-assisted

writing. By integrating scaffolding strategies, AI can be reframed as a thinking partner that supports learner

autonomy, reduces cognitive barriers, and promotes ethical and effective classroom integration (Vygotsky,

1978; Lantolf & Poehner, 2019; Luckin et al., 2023).

writing processes.

The implementation of ScaffoldED AI was carried out in three phases. During the induction phase, students

were introduced to AI tools and given explicit guidelines on ethical and effective use. In the intervention phase,

students engaged in guided writing tasks supported by structured prompts and ChatGPT integration. Finally, in

the post-intervention phase, learners engaged in reflection, peer feedback, and the production of finalised

writing outputs.

The study involved two groups of participants: 32 tertiary STEM students from a public university in Malaysia

and 21 secondary school students from a secondary school in Malaysia, who were purposively selected as they

had been identified as struggling with English proficiency. A mixed-methods research design was employed to

capture both quantitative and qualitative insights. The quantitative strand utilised a questionnaire consisting of

themes.

To capture learner perspectives, a survey was conducted with both tertiary and secondary ESL students. Table

1 presents the comparative results across key areas, including comprehension, efficiency, collaboration, and

language development, highlighting how ScaffoldED AI supported learners in different educational contexts.

The findings from both groups of participants demonstrate that ScaffoldED AI was highly effective in

supporting ESL learners at different educational levels. A large majority of tertiary students (94%) and

secondary students (95%) reported that AI helped them to better understand difficult topics, suggesting that AI-

assisted scaffolding can reduce cognitive barriers to comprehension. Similarly, 88% of tertiary learners and

100% of secondary learners indicated that AI significantly saved them time in analysing large amounts of

information and locating relevant sources, highlighting the efficiency gains associated with structured AI use.

The study also revealed benefits in collaborative and creative aspects of writing. While 47% of tertiary

students and 62% of secondary students found AI useful for presentations and group projects, higher

proportions noted gains in the quality of feedback and suggestions: 47% for tertiary and 76% for secondary

of secondary learners recognising improvements in grammar and vocabulary, and 82% and 95% respectively

acknowledging its role in expanding ideas.

To complement these quantitative findings, the qualitative responses offered valuable insight into how learners

perceived the role of AI in their writing. A recurring theme was the view of AI as a supportive partner in idea

generation and expansion. Several students described how AI gave them a starting point when they struggled

to begin a task, with one noting that “AI is good for expanding our ideas, especially when brainstorming.”

Another student highlighted its motivational effect, explaining that “ChatGPT is less intimidating than starting

In addition, there was a strong recognition that while AI can provide technical support, it cannot replace the

human dimensions of teaching and mentorship. One participant noted, “AI could replace human aspects of

teaching, like empathy and mentorship,” highlighting that the relational and affective roles of teachers remain

central to the learning process. These qualitative findings reveal that students are not only aware of the benefits

AI can bring but are also critically reflective of its limitations, pointing to the necessity of responsible and

scaffolded integration.

These findings highlight the dual role of AI as both a cognitive partner and a sociocultural tool. On the

cognitive level, AI reduced the burden of planning and translating, allowing students to focus more on clarity

structured prompts and teacher mediation ensured that AI became a form of guided interaction within the Zone

drafting, feedback, and reflection foster deeper engagement. At the same time, the different response rates

between tertiary and secondary learners suggest that younger students may benefit more quickly from AI

scaffolding in foundational skills, whereas tertiary learners require support for higher-order tasks such as

to express their personal voice more effectively. At the same time, the role of teacher guidance and structured

For institutional implementation, ScaffoldED AI could be introduced through phased professional development

programmes that train educators to integrate structured prompts into their teaching practice. Institutions may

adopt licensing models to embed AI platforms within existing learning management systems, enabling

streamlined access and analytics. Pilot implementations across multiple faculties or schools could then inform

broader rollouts, ensuring scalability and sustainability.

The authors would like to express their sincere gratitude to the participating students and teachers whose

cooperation made the implementation of ScaffoldED AI possible.