Students' LLM use in CS2 fell into three themes: understanding, assisting code writing, and writing code, with only a few using LLMs to write code¶

Claim · All claims
Evidence · 1 study (2 entries) · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2031: baselines and conditions change

Subclaims¶

q2 i? Iterative qualitative coding of 198 Week 13 reflection responses found students' LLM usage could be categorized into three themes: used to understand, used to assist code writing, and used to write code. → Margaret Ellis 2026 q2 i? Most students stated no LLM was used for the assignment described, and some students sought no assistance at all. → Margaret Ellis 2026 (2)

Evidence¶

Margaret Ellis 2026¶

Margaret Ellis, Nikitha Donekal Chandrashekar, Sehrish Basir Nizamani, Mohammed Farghally, Jake O'Brien, and Naren Ramakrishnan. (2026). Demystify, Use, Reflect, Assess (DURA): An Experience Report on LLM Integration in CS2. arXiv. https://arxiv.org/abs/2606.30908 doi:10.48550/arxiv.2606.30908

q2 · i? · design · r2

Qualitative analysis of Week 13 Problem Solving Reflections: 198 of 442 students consented and submitted valid responses; five co-authors used an iterative inductive-deductive codebook with 32 help-seeking codes. No effect size printed.

"How, when, and why students did use LLMs varied, their usage can be categorized into the themes: used to understand, used to assist code writing, and used to write code."

Margaret Ellis 2026 (2)¶

Margaret Ellis, Nikitha Donekal Chandrashekar, Sehrish Basir Nizamani, Mohammed Farghally, Jake O'Brien, and Naren Ramakrishnan. (2026). Demystify, Use, Reflect, Assess (DURA): An Experience Report on LLM Integration in CS2. arXiv. https://arxiv.org/abs/2606.30908

q2 · i? · qualitative · r2

Same qualitative coding of 198 Week 13 reflections found most respondents stated no LLM was used on the described assignment, and some sought no assistance. Descriptive; no effect size printed.

"Some students did not seek or use any assistance and most stated that no LLM was used for the assignment described."

Discussion¶

Related Claims¶

• Student reflections identify speed and scaffolding as the dominant theme, with verification burden, overreliance worries, and prompt precision also prevalent — related
• Engineering students most often use LLMs for academic writing (27%), conceptual clarification (20%), programming help (17%) and brainstorming (13%) — related
• Students predominantly recommended future students use office hours and formal course staff, with LLMs positioned as a supplement — related
• Students use the integrated AI assistant differently across writing stages: clarification early, verification late — related
• Students' stated pre-interaction intentions only weakly correspond to their enacted help-seeking with an LLM — related