Exploring the impact of artificial intelligence application in personalized learning environments: thematic analysis of undergraduates’ perceptions in China

Exploring the impact of artificial intelligence application in personalized learning environments: thematic analysis of undergraduates’ perceptions in China Wang Xu Zhang Hao Jie

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Exploring the impact of artificial intelligence application in personalized learning environments: thematic analysis of undergraduates’ perceptions in China 2024 Humanities and Social Sciences Communications 10.1057/s41599-024-04168-x wang2024-exploring-the-impact-of-artificial wang2024-exploring-the-impact-of-artificial.txt, the project's staged copy 12e0242c1a7c26b6d79e77f95ae2c25b41ec2ebd open-access education

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them existence Quotation withheld because this record is not cleared for redistribution. undergraduates' perceptions of generative AI in personalized learning Quotation withheld because this record is not cleared for redistribution. observational-study observational-study Quotation withheld because this record is not cleared for redistribution. mechanisms AI supplies instant feedback and flags common mistakes, which triggers student self-reflection and thereby develops cognitive and thinking skills Quotation withheld because this record is not cleared for redistribution. mechanisms the same personalization and tutoring that improves the learning experience invites excessive reliance, which erodes learner autonomy and critical thinking Quotation withheld because this record is not cleared for redistribution. Small sample and purely qualitative data limit generalizability and may carry subjective bias. what this limits: the generalisability of every finding: THE LIMITED SAMPLE SIZE ... AND THE EXCLUSIVE RELIANCE ON QUALITATIVE DATA Quotation withheld because this record is not cleared for redistribution. No long-term or longitudinal evidence about sustained effects of AI use on learning, cognition or motivation. what this limits: any claim about sustained effects: THE STUDY DOES NOT ADDRESS THE LONG-TERM EFFECTS OF AI USAGE IN EDUCATION Quotation withheld because this record is not cleared for redistribution. Three limits are stated together: a small sample, exclusive reliance on qualitative data with subjective bias, and no long-term effects. what this limits: these findings as evidence about AI in personalised learning generally: the authors name a limited sample, exclusive reliance on qualitative data which may introduce subjective biases, and no treatment of long-term effects Quotation withheld because this record is not cleared for redistribution. The central construct is measured through instruments derived from a taxonomy, so the result inherits that taxonomy's division of cognition. what this limits: a claim about cognitive skills in this study: they are measured with tools developed under guidance from Bloom's Taxonomy, which fixes which abilities count Quotation withheld because this record is not cleared for redistribution. The paper describes its own data collection two incompatible ways. Methods: interviews of about 30 minutes, audio-recorded and transcribed verbatim. Data availability: a survey of 10 questions answered as real-time feedback posted directly on a forum. These are different instruments producing different data, and the abstract uses both words in one sentence ('surveyed 48 undergraduates ... through semi-structured interviews'). Nothing in the paper reconciles them. Quotation withheld because this record is not cleared for redistribution. The 90% agreement is computed after the two coders built the themes together: the preliminary codes were 'collaboratively discussed and categorized into potential themes by both the first and second authors', and only then did each researcher independently apply them. The figure measures consistency of application, not independent arrival at the same scheme. Quotation withheld because this record is not cleared for redistribution. The sample is selected on the exposure under study - students who chose to enrol in an 'AI in Academic Writing' course - and the paper offers that selection as what makes the data informative rather than as a threat to it. Perceptions of AI are then reported from a group defined by having opted into using AI. Quotation withheld because this record is not cleared for redistribution. The paper recommends that future work 'employ quantitative methods alongside qualitative approaches to validate observed trends', while its own results are reported almost entirely as counts out of 48 - 35, 29, 25, 24, 22, 21, 18, 15, 10, 9 students and so on. The trends it asks future work to validate quantitatively are already stated as numbers here. Quotation withheld because this record is not cleared for redistribution. A PROCEDURAL SAFEGUARD ON INSTRUMENT DESIGN: each interview question grounded in both a theoretical definition and actual student responses eliminates: construct drift between what the instrument asks and what it claims to measure - a question written only from theory can miss how students actually talk, and one written only from their words can measure something the construct does not name · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. cognitive skills cognitive skills Quotation withheld because this record is not cleared for redistribution. personalized learning personalized learning Quotation withheld because this record is not cleared for redistribution. generative AI generative AI Quotation withheld because this record is not cleared for redistribution. semi-structured interview semi-structured interview Quotation withheld because this record is not cleared for redistribution. convenience sampling convenience sampling Quotation withheld because this record is not cleared for redistribution. thematic analysis thematic analysis Quotation withheld because this record is not cleared for redistribution. motivation motivation Quotation withheld because this record is not cleared for redistribution. cognitive skills - operationalised through Bloom's Taxonomy: memory, attention, comprehension and reasoning, as measured by tools built under that taxonomy's guidance Quotation withheld because this record is not cleared for redistribution. AI assistance - AS EXPERIENCED BY A PARTICIPANT: a failure mode in which attempting to correct the system can INCREASE its misunderstanding rather than reduce it Quotation withheld because this record is not cleared for redistribution. the data - AS METHODS DESCRIBES IT: interviews of about 30 minutes, audio-recorded and transcribed verbatim by transcription software, then reviewed and edited Quotation withheld because this record is not cleared for redistribution. the data - AS DATA AVAILABILITY DESCRIBES IT: a survey of 10 questions on a platform, answered as real-time feedback posted directly on a forum. NOT THE SAME INSTRUMENT as the transcribed 30-minute interviews of the methods section Quotation withheld because this record is not cleared for redistribution. replicates → Mao, Chen, and Liu (2024) Quotation withheld because this record is not cleared for redistribution. replicates → Seo et al. (2021) Quotation withheld because this record is not cleared for redistribution. replicates → Ghimire et al. (2024) Quotation withheld because this record is not cleared for redistribution. builds-on → Vasconcelos et al. (2023); Passi and Vorvoreanu (2022) Quotation withheld because this record is not cleared for redistribution. cognitive skills —operationalised-as→ semi-structured interview cognitive skills semi-structured interview strength of the relation: 0.9 · note: THE SAME SHAPE ON THE OTHER CONSTRUCT: '…generative AI can enhance your higher-order…' thinking and, if so, how AI achieves it. Both of the paper's two constructs are elicited by a question whose second clause presupposes the effect. The construct is not measured; AGREEMENT WITH A PROPOSITION ABOUT THE CONSTRUCT IS. (This span is column-interleaved in the staged PDF text with a neighbouring column so the quote is exact but the paper's line order is not recoverable from it.) Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span is an interview question about cognitive skills; operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. inter-rater agreement inter-rater agreement Quotation withheld because this record is not cleared for redistribution. Thirty-five students said generative AI helped with homework and exam preparation by offering problem-solving strategies, refining articles and creating course content. Quotation withheld because this record is not cleared for redistribution. Forty-two students said AI supplied learning resources and suggestions tailored to their learning behaviour and past performance. Quotation withheld because this record is not cleared for redistribution. Thirty-three students credited generative AI with fostering their creativity by supplying diverse perspectives and innovative solutions. Quotation withheld because this record is not cleared for redistribution. Twenty-two students worried that excessive reliance on AI could weaken their ability to think independently. Quotation withheld because this record is not cleared for redistribution. Twenty-one students said AI-generated content was sometimes erroneous or inaccurate, so they had to verify and filter it themselves. Quotation withheld because this record is not cleared for redistribution. Twenty-five students were worried about data security and privacy because of the volume of personal data AI systems process. Quotation withheld because this record is not cleared for redistribution. Fifteen students said the AI tools failed to fully understand their queries and returned answers below their expectations. Quotation withheld because this record is not cleared for redistribution. personalized learning environment (PLE) personalized learning environment (PLE) Quotation withheld because this record is not cleared for redistribution. personalized learning environment personalized learning environment Quotation withheld because this record is not cleared for redistribution. engagement engagement Quotation withheld because this record is not cleared for redistribution. motivation motivation Quotation withheld because this record is not cleared for redistribution. cognitive skills cognitive skills Quotation withheld because this record is not cleared for redistribution. motivation motivation Quotation withheld because this record is not cleared for redistribution. learning engagement learning engagement Quotation withheld because this record is not cleared for redistribution. 48 — number of undergraduate interviewees in the study Quotation withheld because this record is not cleared for redistribution. 23 males and 25 females, aged between 18 and 21 — sex and age composition of the student cohort Quotation withheld because this record is not cleared for redistribution. over 90% — agreement between the two coders on themes, codes and quotations Quotation withheld because this record is not cleared for redistribution. four hours weekly — minimum required AI-platform use per student per week Quotation withheld because this record is not cleared for redistribution. 35 — students reporting AI helped with homework and exam preparation Quotation withheld because this record is not cleared for redistribution. 42 — students reporting AI-customized resources boosted their enthusiasm Quotation withheld because this record is not cleared for redistribution. 22 — students fearing over-reliance would reduce independent thinking Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1057/s41599-024-04168-x.