AI multimodal feedback is rated higher than educator feedback on specificity, clarity, simplicity, motivation, satisfaction, and lower mental effort, with comparable correctness, trust, and acceptance¶

Claim · All claims
Evidence · 1 study (3 entries) · causal r2 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 3 of 3 entries) · not reviewed by a person

Subclaims¶

q3 i? Participants rated AI feedback significantly higher on specificity, clarity, and simplicity (all p < 0.05). → Chloe Qianhui Zhao 2026 q3 i? Participants reported significantly lower mental effort and higher satisfaction and motivation with AI feedback. → Chloe Qianhui Zhao 2026 (2) q3 i? No significant differences were observed for correctness (p = 0.5378), trust (p = 0.1456), or acceptance (p = 0.2393). → Chloe Qianhui Zhao 2026 (3)

Evidence¶

Chloe Qianhui Zhao 2026¶

Chloe Qianhui Zhao, Jie Cao, Jionghao Lin, and Kenneth R. Koedinger. (2026). LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback. LAK 2026. https://arxiv.org/abs/2601.15280

q3 · i? · causal · r2

Mann–Whitney U tests of post-survey Likert ratings (81 baseline vs. 99 AI valid responses) in RQ3. Table 2 reports specificity U = 2638.00, p = 0.0001; clarity U = 3346.50, p = 0.0199; simplicity U = 3246.00, p = 0.0207; no effect sizes printed.

"AI feedback received significantly higher ratings for specificity, clarity, and simplicity (all𝑝< 0.05), indicat- ing that participants found the AI-facilitated multimodal feedback to be clearer, more understandable, more concise, and more fo- cused than original educator feedback."

Chloe Qianhui Zhao 2026 (2)¶

Chloe Qianhui Zhao, Jie Cao, Jionghao Lin, and Kenneth R. Koedinger. (2026). LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback. LAK 2026. https://arxiv.org/abs/2601.15280

q3 · i? · causal · r2

Post-survey Mann–Whitney tests in RQ3. Mental effort (rated 1–10) was lower for AI feedback and satisfaction (5-point Likert) higher; motivation was also higher (U = 3174.50, p = 0.0146). No effect sizes printed.

"Participants reported significantly lower mental effort when interacting with the AI Multimodal Feedback (𝑈= 5190.50,𝑝= 0.0006**), while their satisfaction scores were also significantly higher for AI feedback compared to original edu- cator feedback (𝑈=3310.50,𝑝=0.0240)."

Chloe Qianhui Zhao 2026 (3)¶

Chloe Qianhui Zhao, Jie Cao, Jionghao Lin, and Kenneth R. Koedinger. (2026). LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback. LAK 2026. https://arxiv.org/abs/2601.15280

q3 · i? · causal · r2

Post-survey comparisons in RQ3; trust (p = 0.1456) and acceptance (p = 0.2393) also showed no significant differences, which the authors note contrasts with prior studies reporting much lower trust in AI feedback.

"No significant difference was observed for correctness (𝑝= 0.5378), suggesting that learners perceived the factual accuracy of AI and original educator feedback to be similar."

Studies bearing on this claim¶

Do learners rate clearer, more specific, more structured feedback as easier to understand and use? This table gathers the studies behind this claim and the 5 narrower claims that bear on it: 6 distinct studies (2 review, 2 qualitative, 1 causal, 1 design), q2–q3, 0 of 6 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Related Claims¶

• Open-ended learner feedback revealed four themes: trust in AI scaffolding, feedback clarity, engagement, and task authenticity — related
• Personalized paths reduced self-reported cognitive load on all six adapted NASA-TLX dimensions with large effect sizes — related
• AI chatbot access lowers teachers' perceived cognitive effort during learning design — 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 — reports the opposite
• Student perceptions of GenAI feedback functionalities diverged, with some finding them redundant when feedback was perceived as clear — related
• K-12 students across five schools valued CyberScholar's detailed, specific, rubric-aligned feedback as helpful for identifying areas to revise — related
• Trust in customized agents was associated with clear, accurate, and contextually aligned feedback — related
• Students perceived instructor feedback as significantly more useful than AI-generated feedback, with no significant differences on fairness, developmental value or encouragement — reports the opposite
• Students' perceptions of mastery, emotions, and satisfaction were similarly high across all three feedback conditions — reports the opposite
• Discrepancies between lenient learner feedback and more critical automated feedback caused tutors confusion, distrust, and demands for rationale — related