The competence paradox: broadly capable LLMs cannot genuinely unknow expert knowledge when simulating novices¶

Claim · All claims
Evidence · 1 study · theoretical r2 · q1
Currency · newest study 2026 · checked against its sources 2026-10-17 (automated, 1 of 1 entry) · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q1 i? LLM-based simulators exhibit a competence paradox rooted in prior-knowledge entanglement: prompted novices still reason with expert priors, so errors resemble superficial deviations from expert reasoning rather than stable, diagnosis-relevant misconceptions. → Zhihao Yuan 2026

Evidence¶

Zhihao Yuan 2026¶

Zhihao Yuan, Yunze Xiao, Ming Li, Weihao Xuan, Richard Tong, Mona Diab, Tom Mitchell. (2026). Towards Valid Student Simulation with Large Language Models. https://arxiv.org/abs/2601.05473 doi:10.48550/arxiv.2601.05473

q1 · i? · theoretical · r2

Theoretical argument in the formalization section: the article argues the competence paradox is rooted in an irreducible prior-knowledge entanglement problem, since mainstream LLMs are trained to be broadly capable and self-correcting. No empirical evaluation is offered; the authors state the claim is not validated through direct pedagogical interaction.

"Unlike human learners, they cannot genuinely “unknow” solution schemas or expert heuristics once internalized, so even when prompted to act like a novice, their latent reasoning trajectories remain shaped by expert priors"

Discussion¶

Related Claims¶

• LLM student simulators exhibit near-zero Selective Flip Score, flipping at similarly high rates under targeted, misaligned, and generic feedback — related
• Stronger problem-solving capability predicts lower misconception faithfulness: flip rate under generic feedback correlates negatively with SFS (Pearson r = -0.94, p = 0.002) — related
• Multi-turn reflective interaction does not restore selective updating and destabilizes weaker simulators — related
• Chain-of-Thought rationale quality shows no direct correlation with the accuracy of simulated student responses — related
• EducaSim's LLM-based speaker oracle distinguishes it from prior simulation tools on multi-agent interaction — related