Students rate LLM-generated response-level feedback favourably, with clarity and perceived accuracy highest and actionability and usefulness lowest¶

Claim · All claims
Evidence · 1 study · associational r1 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Across the five response-level dimensions, student-level means ranged from 4.24 to 4.43, with clarity (M = 4.43) and accuracy (M = 4.40) highest and actionability (M = 4.24) and usefulness (M = 4.24) lowest. → Mendonça PC 2026

Evidence¶

Mendonça PC 2026¶

Mendonça PC, Quintal F, Figueiredo M and Mendonça F (2026) Perceived usefulness and intention to use large language model-generated feedback across three educational levels: a user-centred study in programming. Front. Educ. 11:1934069. https://doi.org/10.3389/feduc.2026.1934069

q3 · i? · associational · r1

Cross-sectional rating study of 144 students across three educational levels who rated 893 response-level LLM feedback instances on five-point Likert items. Descriptive results show "Clarity (M = 4.43) and accuracy (M = 4.40) had the highest means"; favourable ratings (mean ≥4) ranged from 72.2% for actionability to 83.3% for clarity.

"Clarity (M = 4.43) and accuracy (M = 4.40) had the highest means, and actionability (M = 4.24) and usefulness (M = 4.24) had the lowest."

Discussion¶

Related Claims¶

• Students rate LLM-generated consolidated reports favourably, with cognitive load rated lowest — related
• Perceived usefulness of response-level LLM feedback is most strongly associated with actionability and perceived accuracy — related
• Students in the generative AI-supported condition reported favorable perceptions of the instruction, with feedback for correcting mistakes rated highest — related
• Under blind conditions, students rated GenAI feedback more positively than tutor-generated feedback, with a statistically significant difference only on Genuineness favouring GenAI — related
• Differences between human and ChatGPT feedback quality were small, with means within one point and partial eta-squared ranging from .01 to .12 — related
• Students rate human-generated feedback as superior to AI-generated feedback in quality, helpfulness, and clarity, and as more likely to be acted upon — related
• Only perceived accuracy differed across educational contexts, rated lower in CTeSP than in secondary and undergraduate education — related