Human scoring and feedback for HDR takes about 11 minutes 30 seconds per learner, while GPT-based scoring and feedback is nearly instant¶

Claim · All claims
Evidence · 1 study · design r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Human scoring and feedback for HDR takes approximately 11 minutes and 30 seconds per learner; for 100 learners it would take approximately 17 hours and 8 minutes, whereas GPT-based scoring and feedback can be completed almost instantly. → Satoshi Takahashi 2026

Evidence¶

Satoshi Takahashi 2026¶

Satoshi Takahashi, Atsushi Yoshikawa, Megumi Kose, Kenichi Suzuki, Chieko Inoue, Yumi Watanabe, and Mari Sawada. (2026). Automating Constructive Assessment with Large Language Models: Toward Scalable and Repeated Evaluation of Practical Competence. arXiv. https://arxiv.org/abs/2609.25790 doi:10.48550/arxiv.2609.25790

q3 · i? · design · r2

Time analysis (Table 8): five graders spent 138 minutes scoring 117 responses plus 30 minutes consensus; feedback preparation, creation, and re-check took 860 minutes for 117 questions, versus a few seconds for LLM generation.

"when humans provide scoring and feedback for HDR, it takes approximately 11 minutes and 30 seconds per learner. If 100 learners were assessed, it would take approximately 17 hours and 8 minutes."

Discussion¶

Related Claims¶

• Automated HDR problem generation makes repeated and longitudinal assessment feasible by producing structurally equivalent cases with different contexts — related
• Automated marking reduced assessment turnaround from 11.2 days to 0.1 days and enabled intervention students to complete 4.1 times the practice volume of controls — related
• In a real academic pilot, AISSA processed 90 presentations reliably at 1–3 minutes per submission and an estimated cost of $0.06–0.07 USD per evaluation — related
• A page-by-page, rubric-guided multimodal LLM workflow graded an authentic handwritten exam in about three hours at roughly $100 in token costs, versus about $3,500 in TA time — related
• 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 — related
• GPT-4 generated detailed per-response feedback that was useful for identifying scoring errors and could support personalized student feedback — 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 — a narrower finding that bears on this claim
• GPT-generated feedback passes simulated-student helpfulness evaluations only about 35% of the time on real student data — related
• LLMs can deliver scalable, personalized, interactive feedback almost instantly, reducing dependency on teacher availability and supporting self-regulated learning in large courses — a broader claim this one bears on
• Simulated Human-in-the-Loop workflow reduces required full human scoring of written productions by at least 50% — related
• Under a simulated load of 100 concurrent users, Taklif.AI averaged 30 seconds end-to-end response time per request — related
• Early adopters report time savings from ChatGPT automation, including a Wharton professor's exam-writing productivity increase of 100% — related
• LLM-based extraction of three variables from 156 meta-analyses cost about $50 in computing fees, far less than estimated human-assistant labor costs — related