Synthesis matrix example · 8 real papers · generated 2026-07-14
AI writing tools academic writing: a worked synthesis matrix
Every row below comes from a real paper (click any citation to verify at the source). This is exactly what the generator produces — treat it as a first draft to edit, not a finished review.
| Citation | Research question | Method | Sample | Key findings | Limitations |
|---|---|---|---|---|---|
| Dergaa et al. (2023) | What are the benefits, threats, and ethical considerations of using ChatGPT in academic writing? | Quasi-qualitative literature review of Scopus-indexed Q1 journals. | Peer-reviewed scholarly articles. | AI enhances efficiency but poses significant risks to the authenticity and credibility of academic work. | The study relies on existing literature rather than primary empirical data. |
| Khalifa & Albadawy (2024) | In which specific domains does AI support academic writing and research? | Systematic literature review. | 24 studies from PubMed, Embase, and Google Scholar published since 2019. | AI assists in six core domains: idea generation, content structuring, literature synthesis, data analysis, editing, and ethical compliance. | Focuses primarily on biomedical research contexts. |
| Nguyen et al. (2024) | What are the patterns of human-AI interaction during the academic writing process? | AI-driven learning analytics, sequence analysis, and process mining. | 626 recorded activities from 10 doctoral students. | Iterative, highly interactive collaboration with AI leads to better writing performance than linear, supplementary use. | Small sample size limited to doctoral students. |
| Perkins (2023) | What are the academic integrity implications of LLMs in formal assessments? | Conceptual analysis of LLM evolution and educational impact. | Higher Education Institutions (HEIs) and student assessment contexts. | AI can create original, coherent text that evades detection, necessitating updated institutional integrity policies. | Focuses on policy and conceptual frameworks rather than empirical student performance data. |
| Alkaissi & McFarlane (2023) | What are the implications of AI-generated 'hallucinations' in scientific writing? | Case study analysis of ChatGPT-generated medical reports. | Two medical case reports (homocystinuria-associated osteoporosis and Pompe disease). | ChatGPT can produce plausible but factually incorrect information, posing risks to scientific accuracy. | Limited to two specific medical case studies. |
| Roe et al. (2023) | How can AI-powered writing tools be categorized to inform pedagogical and integrity policies? | Conceptual inductive analysis. | Language classroom contexts and existing writing technologies. | Tools are categorized into machine translators, digital writing assistants, and automated paraphrasing tools, highlighting a gap in institutional policy. | Conceptual framework lacks empirical testing in diverse classroom settings. |
| Özfidan et al. (2024) | What are the perceptions of Saudi undergraduates regarding AI tools in academic writing? | Survey and exploratory factor analysis. | 189 Saudi undergraduate students. | Students value AI for idea generation and grammar, but express concerns regarding reliability and over-reliance. | Geographically limited to the Saudi context. |
| Trần (2024) | What are the effects of AI tools on teaching and learning English academic writing skills? | Mixed-methods (tests, questionnaires, and interviews). | 5 teachers and 60 students in Hanoi. | AI tools positively impact cohesion, coherence, and lexical resources at discourse and sentence levels. | Small sample size restricted to specific language centers in Vietnam. |
Themes across these papers
Academic Integrity and EthicsDergaa et al. (2023) · Perkins (2023) · Alkaissi & McFarlane (2023) · Roe et al. (2023) · Özfidan et al. (2024)
These papers address the tension between AI-driven efficiency and the risks of plagiarism, hallucination, and policy inadequacy.
Pedagogical Integration and Student PerformanceKhalifa & Albadawy (2024) · Nguyen et al. (2024) · Özfidan et al. (2024) · Trần (2024)
These papers explore how AI tools are practically applied in writing tasks and the resulting impact on student writing quality.
Research gaps identified
- →Lack of longitudinal studies tracking the long-term impact of AI-assisted writing on student critical thinking development.
- →Insufficient empirical evidence on how AI tools affect the writing performance of non-native English speakers across diverse cultural contexts.
- →Absence of standardized frameworks for institutional AI-integrity policies that balance academic rigor with technological accessibility.
- →Limited research on the specific 'hallucination' rates of AI tools across different academic disciplines beyond medicine.
Suggested reading order
- 1.Roe et al. (2023) — Provides a foundational taxonomy of AI tools to establish terminology.
- 2.Khalifa & Albadawy (2024) — Offers a broad overview of AI's functional domains in research.
- 3.Nguyen et al. (2024) — Examines the nuanced human-AI interaction patterns.
- 4.Trần (2024) — Shows practical application in language learning.
- 5.Özfidan et al. (2024) — Provides student-centered perspectives on usage.
- 6.Dergaa et al. (2023) — Introduces the core ethical and authenticity debates.
- 7.Perkins (2023) — Discusses the institutional policy implications of these debates.
- 8.Alkaissi & McFarlane (2023) — Highlights the critical technical risk of AI hallucinations.