Systems capable of performing tasks requiring human-like intelligence—perception, reasoning, learning, decision-making—by analyzing data to generate predictions, decisions, content, or recommendations.

Operationally: Algorithm-based tools that generate and summarize texts, translate, answer research questions, perform statistical analysis of data collection tools, extract and synthesize results, using applications such as interactive chatbots, text summarization, and consensus tools—helping female postgraduate students at the University of Hail improve accuracy, speed of information access, and research skills.

A form of scholarly writing aimed at presenting ideas or studies in an organized, methodical manner per academic standards—clear, precise, objective style; proper source documentation; methodological organization; analysis and synthesis; and adherence to linguistic accuracy and academic integrity.

Operationally: The ability of female postgraduate students at the University of Hail to write scholarly articles, reports, and research papers correctly and systematically across criteria such as text structure, coherence, style, linguistic accuracy, citation integrity, and formatting.

Structured efforts using the scientific method to understand, explain, and relate phenomena, collecting and analyzing data to generate knowledge.

Operationally: The organized steps and rules followed by female postgraduate students at the University of Hail—using AI applications—to study phenomena, generate concepts, collect and analyze data, aiming to improve academic writing.

AI is the development of computer systems capable of tasks typically requiring human intelligence—perception, speech recognition, learning, decision-making, translation. SDAIA defines AI applications as systems that make predictions, generate content, provide recommendations, or take decisions autonomously at varying levels.

AI applications enhance researchers’ capabilities by precisely analyzing problems, generating academic content, extracting knowledge from texts, and improving result accuracy via unbiased data processing. They provide reliable outputs and recommendations, deepen idea development, and improve understanding and analysis of research problems.

Academic writing is foundational to research communication, requiring clear ideas, precise language, organization, critical thinking, and adherence to scholarly standards. Skills include structuring introductions and aims, reviewing literature, aligning methods, analyzing and interpreting results, crafting conclusions/recommendations, organizing ideas, appropriate wording, grammatical accuracy, citation integrity, and formatting.

AI ethics encompass values and principles guiding ethical behavior in developing and using AI: integrity and fairness (mitigating bias, aligning with cultural and human rights values), privacy and security (protecting personal data, cybersecurity), humanity (justice, respect for rights), reliability and safety (compliance with specifications), transparency and explainability (building trust), and accountability (developers responsible for harms). Academic use of AI must align with these principles.

Key obstacles: limited awareness of AI’s importance; financial cost; lack of secure systems protecting data; absence of usage guidelines; insufficient training time and courses; weak technical/digital skills; limited technical support.

Helhol (2025) found AI applications improved academic writing skills among Arabic learners (quasi-experimental). Al-Ghamdi & Al-Furani (2024) reported positive student views on an AI app (“Qalam”) improving grammar and organization. Chen (2023) noted low, sometimes unethical AI use and inadequate citation sourcing. Zhai (2022) showed ChatGPT could produce accurate research papers quickly, recommending AI as assistive tools.

Al-Shahrani (2025) found low use of generative AI for research skills and high obstacles among Saudi grad students. Al Dawood & Al-Fuhaed (2025) found low AI use and high perceived challenges among education researchers. McGrath et al. (2023) identified faculty concerns: lack of knowledge, materials, and practices for AI. Titko et al. (2023) found faculty across 10 European countries stressed the need for AI guidelines and had positive attitudes toward AI for researchers.

Studies vary in focus (writing skills, research skills, faculty attitudes), mostly descriptive, with samples across faculty, learners, and documents. This study aligns with prior work targeting graduate students and uses a questionnaire. Prior literature informed this study’s problem, objectives, significance, and instrument design.

A descriptive analytical approach was used to describe reality quantitatively and qualitatively and to characterize the phenomenon.

Population: all female postgraduate (Master’s) students at the University of Hail in semester 1, 1447 AH. Sample: the full population, n=352. Table 1 shows distribution by specialization and number of technical training courses.

A questionnaire was developed based on prior studies (e.g., Al-Sayyad & Al-Salam, 2023; Zaabta & Sabbagh, 2023) and refined by experts. The final instrument has 26 items across two axes: (1) Use of AI applications for developing academic writing in scientific research (16 items); (2) Obstacles to using AI applications for developing academic writing in scientific research (10 items). Personal data: college/department (specialization) and number of technical training courses. A 5-point Likert scale was used (1=Strongly Disagree to 5=Strongly Agree).

Content validity was assessed by expert review; items were revised/added accordingly. Internal consistency validity: Pearson correlations between items and their axis, and between axes and total score, were high and significant (p<0.01). Axis 1–total r=0.858; Axis 2–total r=0.816, indicating good internal structure.

Cronbach’s alpha was calculated (R software): Axis 1 = 0.91; Axis 2 = 0.77; whole questionnaire = 0.89, indicating high internal consistency.

The questionnaire was administered electronically via: https://forms.gle/cFVE5LwHVqNQLz5j6.

Descriptive statistics (means, SDs, weighted means, percentages) to assess item/axis levels; Cronbach’s alpha; Pearson correlations; independent-samples t-test (by specialization); one-way ANOVA with Scheffé post-hoc (by number of training courses). Likert levels: Very High (4.20–5.00), High (3.40–4.19), Moderate (2.60–3.39), Low (1.80–2.59), Very Low (1.00–1.79).

Table 5 shows means for Axis 1 (use). Overall mean = 3.383 (SD 0.722; 68%, moderate).

Highest use: instant translation/detection (mean 3.79), secure cloud storage (3.70), linguistic error detection (3.59). Lowest: obtaining literature (2.88), using chatbots for research questions (3.04), audio-to-text (3.08). Usage concentrates on quick, direct support tasks; advanced research tasks remain limited.

Table 6 shows Axis 2 (obstacles). Overall mean = 3.644 (SD 0.492; 73%, high). Questionnaire total mean = 3.95 (79%, high).

Top obstacles: lack of guidelines (4.36), weak technical skills (4.32), lack of time (4.31), heavy workload (4.23), weak technical support (4.06). Technical/institutional gaps dominate.