Frontier LLMs barely exceed a trivial baseline on knowledge tracing, with the best model surpassing always-predict-correct by only about five percentage points¶
Claim · All claims
Evidence · 1 study · designr2·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¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? On question-level knowledge tracing over 500 sampled students, the best models (GPT-OSS-120B at 56.2%, Qwen3-Next-80B-Thinking at 55.8%) beat the 51.3% trivial baseline by only around five percentage points, while Llama-3.3-70B reached 52.3%. → Eamon Worden 2026
Evidence¶
Eamon Worden 2026¶
Eamon Worden, Cristina Heffernan, Neil Heffernan, and Shashank Sonkar. (2026). FoundationalASSIST: An Educational Dataset for Foundational Knowledge Tracing and Pedagogical Grounding of LLMs. https://arxiv.org/abs/2602.00070 doi:10.48550/arxiv.2602.00070
q2 · i? · design · r2
Zero-shot LLM evaluation of four models on question-level binary correctness prediction (approximately 27,000 predictions per model, e.g. N 27,715), described in Section 6.1.1. The authors report that the models' gains over the 51.3% baseline were limited, quoting that they "reach 56.2% and 55.8%, respectively, beating the baseline by just around 5 percentage points".
"The best-performing models, GPT-OSS-120B and Qwen3-Next-80B-Thinking, reach 56.2% and 55.8%, respectively, beating the baseline by just around 5 percentage points."
Discussion¶
Related Claims¶
- LLMKT outperforms existing knowledge tracing methods at predicting student turn correctness in the CoMTA and MathDial tutoring dialogue datasets, and generally outperforms DKT-Sem. — the general claim this finding bears on, with the studies behind it
- In system cold-start scenarios, CLST outperformed every baseline KT model across all five datasets, with the largest gains at 8 training students — related
- Representing exercises by KC name descriptions outperformed ID-based representation when aligning an LLM to knowledge tracing — related
- Knowledge tracing performance on tutoring dialogues is relatively low, with a maximum of around 76% AUC, which the authors take to show dialogueKT is a challenging task. — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
llms-barely-exceed-trivial-baseline-kt |
title |
Frontier LLMs barely exceed a trivial baseline on knowledge tracing, with the best model surpassing always-predict-correct by only about five percentage points |
description |
Frontier LLMs barely exceed a trivial baseline on knowledge tracing, with the best model surpassing always-predict-correct by only about five percentage points |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-12 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — eamon-worden-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.