Studying the effect of AI Code Generators on Supporting Novice Learners in Introductory Programming (CHI 2023)
Lower secondary (grades 7–9) Upper secondary AI and learning
Links to claims
Limits (1)
| Claim | Basis |
|---|---|
| C2 | Answer-giving code generator during practice did not lower one-week unaided retention in 10-17-year-olds; design had forced manual follow-up and output was a worked example |
Review state: first-pass draft, pending human review.
Source note
Syntes: Studying the effect of AI Code Generators on Supporting Novice Learners in Introductory Programming (CHI 2023) (2023)
Källa: 10.1145/3544548.3580919 (arXiv:2302.07427)
Verified content
- Controlled experiment, 69 novices aged 10-17 (coding camps, Canada), matched-groups assignment; 45 Python code-authoring tasks with or without OpenAI Codex, each followed by a manual code-modification task without Codex.
- Training: Codex raised authoring completion 1.15x and scores 1.8x; manual modification performance not lower.
- Immediate post-test (no Codex, next day): authoring 61.3% vs 62.9% (p = .838, d = 0.05); modifying 59.7% vs 59.3% (p = .953).
- Retention post-test one week later (no Codex): authoring 59.1% vs 49.8% (p = .262, d = 0.28); modifying 47.5% vs 34.8% (p = .092, d = 0.41); no significant differences; higher-prior-knowledge learners did significantly better with prior Codex access.
- Limits: small sample, AI limited to authoring tasks with a forced manual modification step after each, code output is itself a worked example of the target skill.
Metadata as recorded
- ID
- AI-035
- Tier
- RCT
- Titel
- Studying the effect of AI Code Generators on Supporting Novice Learners in Introductory Programming (CHI 2023)
- År
- 2023
- Källa/DOI
- 10.1145/3544548.3580919 (arXiv:2302.07427)
- Typ
- peer-reviewed conference paper
- Åldersgrupp
- grundskola-hog;gymnasium
- Teman
- ai-och-larande
- Hämtad
- 2026-10-16
- Full text (fetched 2026-10-16, legal open-access copy, private)
- fulltext.pdf + — via unpaywall: https://arxiv.org/pdf/2302.07427
The original files (PDF or abstract) are kept in the project archive and are not published here.