How are Kuwaiti health students using AI to write in English?
In a peer-reviewed survey of 241 nursing and health-sciences undergraduates, 94.6% reported previous generative-AI use and 83.4% used it for academic work at least sometimes. Students also reported accuracy and dependency concerns, while the single-institution self-report design cannot show that AI improved writing or learning.
Editorial responsibility: The Impact of AI Editorial Desk · Report a factual concern
At a glance
- 1The convenience sample included 241 undergraduates at one Kuwaiti institution: 156 nursing students and 85 health-sciences students.
- 2Previous generative-AI use was reported by 228 students (94.6%); 201 (83.4%) said they used it for academic work sometimes, often or always.
- 3The study measured perceptions rather than writing quality: 72.4% agreed that AI information may be inaccurate and 63.4% worried about dependency.
Research topic
How nursing and health-sciences students at one Kuwaiti public institution report using generative AI for English academic writing
The answer: students used AI as language support and a shortcut through workload
Generative AI was already routine for many of the 241 Kuwaiti nursing and health-sciences students surveyed. Two hundred and twenty-eight participants, or 94.6%, reported having used a generative-AI tool. For academic work, 44.0% said sometimes, 27.8% often and 11.6% always—together 83.4%. Students described using the tools across the writing process, from ideas and outlines to grammar, vocabulary, paraphrasing, summaries and revision.
The strongest pattern was support for writing in English. Among valid item responses, 72.5% reported using AI often or always for grammar, 73.1% for vocabulary and 75.3% when they had difficulty expressing ideas in English. Seventy-three per cent agreed that AI helped them complete writing tasks more efficiently. These are reports of use and perceived help, not tests of whether a paper became more accurate, whether vocabulary was retained or whether students could later write unaided.[1][2]
The Impact Brief · Free
Follow the evidence in education.
Get a five-minute weekday briefing on what changed, why it matters and where the evidence comes from. Choose the topics you care about.
The study combined a survey with short Arabic responses
The research took place at the Public Authority for Applied Education and Training, where the Colleges of Nursing and Health Sciences teach primarily in English. The convenience sample contained 156 nursing students and 85 health-sciences students. There were 166 men and 75 women; 44.4% were in their second year. Instructors distributed the SurveyMonkey link through Microsoft Teams after the researcher visited participating classes in March 2026.
A researcher-developed questionnaire contained demographic items, eight AI-use items, nine perceived-benefit items, eight concern items and six open questions. Two specialists reviewed it, and a 23-student pilot preceded the study. Internal-consistency coefficients were 0.85 for use, 0.90 for benefits and 0.77 for concerns. Those values show that items within each scale correlated; they do not by themselves establish that the instrument measures actual competence, ethical practice or learning.[1][2]
Reported benefits came with equally visible cautions
Seventy-one per cent of valid respondents agreed that AI helped them write more effectively in English, 62.3% that it increased writing confidence and 61.6% that it reduced writing anxiety. Frequent reported uses included generating ideas (65.3%) and summarising academic materials (66.7%). For multilingual students facing disciplinary content and English academic conventions at the same time, immediate language help may feel like useful scaffolding.
The same students did not describe the tools as reliably safe. Of 225 who answered the accuracy item, 72.4% agreed that generated information may contain errors. Of 227, 63.4% worried that students could become overdependent, and 63.9% worried that overreliance could reduce critical thinking. Sixty-six per cent agreed that clearer guidance was needed. This combination is more informative than a simple adoption rate: students valued assistance while recognising that fluent text can be wrong and that convenience may displace practice.[1][2]
The qualitative evidence adds context but not independent confirmation
Between 193 and 203 students answered each of the six open questions. The author used reflexive thematic analysis and identified four connected themes: language facilitation, cognitive and academic efficiency, confidence, and negotiation of responsible use. Responses were written in Arabic and translated into English by the bilingual author, who repeatedly checked them against the originals. That design allowed students to respond in a comfortable language.
The average open response was only 13 to 16 words, however, and one researcher collected, translated and interpreted the material. The prompts explicitly mentioned possible benefits and risks, which may have guided what students discussed. The qualitative strand helps explain the survey patterns but is not an independent behavioural check. It cannot verify whether students disclosed AI in assignments, caught inaccurate outputs or used the tools as a complement rather than a substitute.[2]
What teachers in health programmes can do now
Teachers can replace abstract rules with task-level permission and verification. A writing assignment might allow brainstorming and grammar feedback but require students to preserve drafts, identify substantive AI changes, check every factual claim against an original source and explain a clinical concept without assistance. In health education, generated wording deserves particular scrutiny because confident errors can migrate from coursework into professional habits.
Support should also recognise the language burden that drives adoption. Prohibiting AI without offering writing instruction may remove a tool while leaving the underlying difficulty untouched. Institutions can provide academic-English support, teach source tracing and distinguish proofreading from authorship. Assessment can include oral defence, in-class reasoning or revision commentary so students demonstrate understanding rather than merely submit polished prose.[1][2]
Why the percentages are not a national or causal result
Participants came from two colleges at one public institution and were recruited from accessible classes. The exact number receiving the link was unknown, so the response rate cannot be calculated. A predominantly male sample and missing answers on some items further narrow transferability. The survey was collected in March 2026, while products and institutional rules may change quickly.
No assignment, language test, usage log or academic-integrity record was examined. A student who believes AI improves writing may be correct about fluency yet wrong about factual quality or long-term learning. Cross-sectional self-report cannot determine whether AI caused confidence, anxiety reduction, dependency or lower critical thinking. Nor can it show whether nursing and health-sciences students behave differently from peers in other disciplines or institutions.[1][2]
What would change the assessment
Confidence would rise with representative sampling across Kuwaiti and Gulf institutions, clearer invitation denominators and validated measures that distinguish language correction, idea support and authorship. Longitudinal studies should compare independent writing before and after sustained use, while controlled teaching trials could test whether verification and disclosure instruction improves accuracy without removing useful language support.
The present evidence supports a practical conclusion: AI use is common enough that health educators need explicit, discipline-specific teaching now. It does not support claims that generative AI has improved learning, that most students misuse it, or that reported confidence will transfer to clinical communication. The author reported no funding and no competing interests. Anonymised data may be requested subject to institutional and ethical limits.[1][2]
What this means for people
- Students may gain immediate language support, but polished AI-assisted prose should not be mistaken for independent understanding.
- Teachers need to set permissions by task and teach verification, especially where health information can affect future practice.
- Institutions should pair AI rules with academic-English support so multilingual students are not left with the same unmet need.
Global context
Most higher-education AI evidence has come from Western or East Asian settings. This Kuwaiti study adds a Gulf, English-medium health-education perspective and shows why language context matters. Its practical questions are global—how to preserve authorship, verification and learning while students already use AI—but policies should be tested locally rather than extrapolated from one institution's percentages.
What the evidence does not yet show
- The convenience sample came from two colleges at one Kuwaiti public institution; the number invited was unknown, so no response rate could be calculated.
- Use, benefit and concern were self-reported; there were no writing assessments, usage logs, source checks or learning outcomes.
- The researcher-developed instrument had limited validation beyond expert review, a small pilot and internal-consistency estimates.
- One researcher translated and interpreted brief Arabic open responses, and the question prompts may have shaped the themes.
- The March 2026 cross-section cannot show causal effects or track fast-changing tools and policies.
What to watch next
- Longitudinal evidence on independent English writing, disciplinary understanding and retention
- Observed rather than reported use, with privacy-protecting consent and clear task categories
- Controlled evaluations of verification, disclosure and AI-literacy teaching
- Comparable sampling across Kuwait, the Gulf and different health disciplines
- Whether students can identify fabricated sources and clinically consequential errors
Living evidence record
Impact record IAI-1TXSBYY
Evidence stage
Studied
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Present
Record status
Monitoring
Last checked
13 October 2026
Source trail
2 direct sources across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Related-source reporting disclosure
This record analyses 2 linked source records around the same underlying development. The extra records add method, date or context, but they do not by themselves constitute independent replication of every performance claim or predicted outcome.
Evidence trail
Sources used for this report
Links checked 13 October 2026
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Continue the story
Related reporting
Education
How are Australian engineering students actually using generative AI?
A peer-reviewed survey of 131 students at two Australian universities found broad but uneven use: coding was reported by 46.6% of one cohort and 7.1% of the other. Because the cohorts differed in size, year level and location, the study cannot identify which factor caused the gap.
7 min · 2 sources
Education
Does AI help students design climate solutions?
In a peer-reviewed UAE study, 69 students reported greater creativity and agency when using AI, but the research measured perceptions in one course—not learning gains, stronger climate solutions or real-world environmental impact.
7 min · 1 source
Education
Did ChatGPT improve writing scores in a 16-week Taiwan course?
Twelve university students used ChatGPT throughout a five-stage inquiry-writing course. Their closed-book writing scores rose by 2.79 points, but the change was not statistically significant and the study had no comparison group.
8 min · 1 source
Reader commentary
Add evidence, experience or a question
No account is required. Reader notes are published after a brief civility, relevance and safety check; disagreement is welcome.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.