DeepTutor's interactive tutoring gains are stable across five university-level domains, varying only 0.16 points in overall quality¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Decomposition of DeepTutor's interactive performance by discipline shows overall quality varies by only 0.16 points across the five domains, with Source Faithfulness the most domain-sensitive metric. → Bingxi Zhao 2026

Evidence¶

Bingxi Zhao 2026¶

Bingxi Zhao, Jiahao Zhang, Xubin Ren, Zirui Guo, Tianzhe Chu, Yi Ma, Chao Huang. (2026). DeepTutor: Towards Agentic Personalized Tutoring. Technical Report, The University of Hong Kong. https://arxiv.org/abs/2604.26962 doi:10.48550/arxiv.2604.26962

q2 · i? · design · r2

Cross-domain decomposition of the first-person interactive simulation (Figure 6) across five disciplines with different discourse structures. The article reports the 0.16-point overall span and a 0.36-point average metric span; these are descriptive scores with no standardized effect size.

"Overall quality varies by only 0.16 points across the five domains, indicating that the gains are not driven by a single discipline. At the metric level, the average cross-domain span is 0.36 points, with the largest fluctuation appearing inSource Faithfulness"

Discussion¶

Related Claims¶

• DeepTutor improves overall first-person interactive tutoring quality by 10.76% over a Naive Tutor baseline on TutorBench — related
• Static Knowledge Grounding and Dynamic Personal Memory are complementary: removing both yields the largest quality degradation — related
• Human raters' preferences align with the LLM judge on TutorBench (Pearson r = 0.82 across metric-level win rates) — related
• LLM teaching effectiveness varies substantially across subject domains, strongest where knowledge points apply directly and weakest in physics and chemistry — related