AI evidence extraction

UK healthcare educators’ experiences and perceptions of essay assessments in the generative AI era

Benard Ohene Botwe, Cletus Amedu, M Ngo · Discover Education · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

1/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1007/s44217-026-01826-x

Methodology & findings

Study design

Exploratory descriptive qualitative design using semi-structured interviews (n=10 healthcare academics) with thematic analysis conducted using NVIVO version-14

Sample

N = 10, 2 groups

Primary method

Thematic analysis using NVivo version-14 software for qualitative data analysis.

Main result

The study found that "Participants reported widespread evidence of AI-generated content in student submissions. They also indicated relying largely on intuitive detection methods such as identifying 'red flags' including suspicious references, shallow content, unusual language patterns, and fabricated elements, among others. However, their inability to definitively confirm AI use created uncertainty, frustration, and ethical discomfort, undermining trust between faculty and students and threatening academic fairness."

Reports effect sizes.

Research paradigm

Interpretivism/Qualitative

Author conclusions

The authors conclude that "healthcare educators continue to observe what they perceive as irresponsible use of GenAI in essay-based assessments, while facing a complex 'detection dilemma.'" They further recommend that "higher education institutions develop clear strategies to address the irresponsible use of AI in assessment and empower educators to detect AI use in essay-based assessments."

Risk of bias

Selection bias: Small sample size (n=10) from specific UK healthcare disciplines may not be representative of all healthcare educators or educational contexts; Interviewer bias: Semi-structured interviews are subject to interviewer interpretation and respondent social desirability bias; Confirmation bias: Educators' preexisting concerns about GenAI may influence their perception of suspicious content; Limited generalizability: Sample limited to UK healthcare educators from specific disciplines (Radiography, Midwifery, Speech and Language Therapy, Nursing); Selection bias: Small sample size (n=10) may not be representative of all UK healthcare educators; Selection bias: Participants recruited from specific disciplines (Radiography, Midwifery, Speech and Language Therapy, Nursing); Interviewer bias: Semi-structured interviews subject to interviewer effects and interpretation; Confirmation bias: Researchers and participants may selectively notice 'red flags' confirming AI use suspicions; Small sample size (n=10) may limit generalizability; Selection bias: participants self-selected from specific healthcare disciplines; Interviewer bias: semi-structured interviews subject to interpretation; Social desirability bias: participants may report idealized perceptions rather than actual practices

Open questions raised

The authors identify limited research on how UK educators perceive and experience GenAI use in traditional essay-type assessments, stating "With limited research on this topic, questions arise about how UK educators perceive and experience the use of GenAI in traditional essay-type assessments (TETAs)."
Limited prior research on how UK educators perceive and experience GenAI use in traditional essay-type assessments (TETAs); need for clearer institutional strategies and educator training on AI detection
The abstract indicates that "With limited research on this topic, questions arise about how UK educators perceive and experience the use of GenAI in traditional essay-type assessments (TETAs)," suggesting a gap in understanding educator experiences with GenAI in assessment contexts.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 64%

Explore related topics

Responsible AI Use Research Integrity Qualitative