The AI feedback system returns responses in under a second for MCQs and about six seconds for open-ended questions, and its reusable prompt is roughly 2.77 times more time-efficient to prepare than equivalent educator text¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q1–q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? Median system response time was 0.299 seconds for MCQs and 6.23 seconds for open-ended questions. → Chloe Qianhui Zhao 2026 q1 i? By word-count proxy, the system is approximately 2.77 times more time-efficient than authoring equivalent human text feedback. → Chloe Qianhui Zhao 2026 (2)

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 doi:10.48550/arxiv.2601.15280

q2 · i? · causal · r2

System response-time calculation in RQ1.2, covering total time to return structured text feedback and the slide reference for each question type. Descriptive statistics only; no comparison test.

"For multiple choice questions, the median of system response time is 0.299 seconds (𝑀𝑒𝑎𝑛= 0.700,𝑆𝐷= 1.457); for open-ended questions, the median of system response time is 6.23 seconds (𝑀𝑒𝑎𝑛=7.210,𝑆𝐷=3.389)."

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

q1 · i? · causal · r2

A proxy comparison of word counts: the prompt template contained 356 words versus 985 words of human-authored feedback across 8 MCQs (26 options) and 5 OEQs. The authors state educator time writing the feedback was not recorded, so this is an assumption-based estimate.

"Assuming com- parable effort per word for preparing a set of reusable feedback, the system is approximately985/356≈ 2.77× more time-efficient than authoring equivalent human text feedback."

Discussion¶

Related Claims¶

• Human scoring and feedback for HDR takes about 11 minutes 30 seconds per learner, while GPT-based scoring and feedback is nearly instant — related
• On the live single-node deployment, short-answer rounds take ≈32 s (≈$0.015/round) while multiple-choice rounds are ≈4× faster (≈7.6 s, ≈$0.005/round) because grading is deterministic — related
• Feedback type shifts engagement by question type: educator feedback yields more MCQ submissions while AI feedback trends toward more open-ended revisions, with no overall difference — related
• Prebuilt quiz banks of 435 questions across 65 modules decouple practice from live generation, with warm practice-request medians of 4.8 and 5.8 ms — related
• LLM-generated retrieval practice was practical to deploy and well received, but high correctness rates (about 70% of items at p=1.0) restricted score variance — related
• Under a simulated load of 100 concurrent users, Taklif.AI averaged 30 seconds end-to-end response time per request — related
• LearnLens reduces teacher feedback time, with half of teachers previously spending 10–30 minutes per assignment versus a median below 5 minutes with the system — related
• LearnLens's modular pipeline outperforms direct-prompt baselines on scoring quality while matching latency and cost on 100 authentic student answers — related
• ConnectED reduces 45-minute lesson preparation time to approximately 15–20 minutes in-system, versus 3–4 hours for manual preparation — related