LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback

Synthesis: The authors built a real-time AI-facilitated multimodal feedback system integrating structured text, relevant slide references, and streaming AI audio narration. In a crowdsourced experiment, they compared it against fixed "business-as-usual" educator feedback.

Core Finding

The System

The multimodal feedback system combines three channels:

1. Structured textual explanations — targeted, specific feedback on student answers
2. Dynamic slide references — retrieval of the most relevant lecture slide for context
3. Streaming AI audio narration — spoken explanation to complement text

This multimodal approach aims to provide timely, targeted feedback at scale — something that remains a major challenge in education.

Results

Learning Effectiveness

• Equivalent Learning Gains between AI multimodal feedback and educator feedback
• No significant difference in post-test performance

Student Perceptions (AI significantly better)

AI feedback matched educator feedback on correctness, trust, and acceptance — but outperformed on every experiential dimension.

Behavioral Engagement Patterns

• Multiple-choice questions: Educator feedback encouraged more total submissions (students kept trying)
• Open-ended questions: AI feedback lowered revision barriers — targeted suggestions promoted iterative improvement

Significance

This is a strong result for AI feedback systems:

• Equivalence on learning is the bar most systems fail to clear — AI matched human educators
• Superior Student Experience across 6 dimensions suggests AI can surpass humans on consistency, specificity, and clarity
• Scalability: The system can provide real-time, context-aware support without instructor availability constraints
• Question-type effects: The engagement pattern differences suggest adaptive strategies — AI may be better for open-ended work while human-like interaction helps for multiple-choice

Methodological Notes

• Online crowdsourcing experiment — participants recruited via a platform, not classroom students
• Compared against fixed educator feedback (not live, not adaptive) — the AI system's adaptivity may partly explain its perceptual advantage
• Single-session design — long-term effects unknown

Open Questions

• Would results hold in real classroom settings with live educator feedback rather than fixed, pre-written feedback?
• Does the advantage persist over multiple sessions, or is there a novelty effect?
• How does each modality (text vs. slides vs. audio) contribute to the overall effect?
• Can Multimodal AI feedback reduce the negative behavioral pathway identified in sequenced feedback studies (fewer resubmissions)?

What this means for practice

• Learners. Use AI feedback as a working explanation rather than a verdict: in this experiment the AI condition matched educator feedback on learning after controlling for pre-test scores (𝐹 = 0.91, partial 𝜂² = 0.0047) and was rated clearer, more specific, and less cognitively demanding.
• Learners. Read the linked slide and play the audio narration alongside the text instead of skimming the text alone — the three channels are designed to carry context that any one of them omits.
• Instructors. Keep live human feedback where it earns its cost. Educator feedback drove more total submissions on multiple-choice questions, while AI feedback lowered the barrier to revising open-ended answers.
• Instructors. Do not assume AI feedback is the weaker option: it was rated equivalent on correctness, trust, and acceptance and better on all six experiential dimensions, including reduced cognitive load.
• Target response latency that allows revision inside the same session — the system returned multiple-choice feedback in a median of 0.299 seconds and open-ended feedback in 6.23 seconds.

Limitations

• The study was a single one-hour online session with 197 Prolific-recruited U.S. university students (87 in the business-as-usual group, 110 in the AI group), so it cannot address knowledge retention, delayed testing, or transfer.
• The comparison condition used fixed, pre-authored educator feedback rather than live or adaptive human feedback, which may account for part of the AI system's perceptual advantage.
• The two conditions differ on several confounded factors at once, including feedback source and slide presentation style, so the study cannot separate their individual contributions to learning.
• Cognitive load was captured only as an overall perceived-load measure, which cannot distinguish a reduction in extraneous load from an increase in generative load.

Citation

Zhao, C. Q., Cao, J., Lin, J., & Koedinger, K. R. (2026). LLM-based Multimodal Feedback Produces Equivalent Learning and Better Student Perceptions than Educator Feedback. Accepted to LAK 2026.