Across 28 project contexts, human-in-the-loop and independent peer review identified no conceptual or mathematical hallucinations, and Python visualization scripts executed without compilation errors on the zero-shot attempt.¶

Claim · All claims
Evidence · 1 study (2 entries) · design r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? No conceptual or mathematical hallucinations were identified during human-in-the-loop review across the 28 project contexts. → Henrique Mohallem Paiva 2026 q2 i? Python scripts generated for data visualization invariably executed without compilation errors on the zero-shot attempt. → Henrique Mohallem Paiva 2026 (2)

Evidence¶

Henrique Mohallem Paiva 2026¶

Henrique Mohallem Paiva. (2026). Curriculum as Code: An AI-Assisted Architecture for Instructional Design in STEM Education. https://arxiv.org/abs/2608.07364 doi:10.48550/arxiv.2608.07364

q2 · i? · design · r2

One-year observational validation reporting review outcomes across 28 project contexts; the article states "no conceptual or mathematical hallucinations were identified" during human-in-the-loop review. No error-rate statistics are printed.

"As a direct result of this architectural constraint, no conceptual or mathematical hallucinations were identified during the human-in-the-loop review across the 28 project contexts."

Henrique Mohallem Paiva 2026 (2)¶

Henrique Mohallem Paiva. (2026). Curriculum as Code: An AI-Assisted Architecture for Instructional Design in STEM Education. https://arxiv.org/abs/2608.07364

q2 · i? · design · r2

Technical fidelity evaluation within the same one-year validation: constrained by Phase 3 parameterization, generated Python scripts "invariably executed without compilation errors on the zero-shot attempt". LaTeX showed only minor syntactic anomalies such as mismatched bracket closures.

"Constrained by the strict parameterization enforced during Phase 3, the Python scripts generated for data visualization invariably executed without compilation errors on the zero-shot attempt, successfully outputting publication-quality figures (e.g., adhering todpi=300and specific library constraints)."

Discussion¶

Related Claims¶

• All AI-generated outputs in the project underwent mandatory human-in-the-loop review, positioning AI as augmentative rather than substitutive — related
• Phase VI AI-enabled systems face six documented challenges: explainability, hallucination risk, prompt sensitivity, computational cost, validation complexity, and limited large-scale evidence — related
• LLM-assisted inductive qualitative coding carries risks of superficial themes, broad or redundant codes, and hallucinated interpretations, so LLMs should augment rather than replace human researchers — related
• The v6m workflow produced compilations passing PDF/UA-2 and ISO 32005 validation in 16 of 17 successive transcription tests across 16 videos — related
• Recurring risks of GenAI in STEAM education include hallucination, bias, superficial completion strategies, and compromised assessment validity — reports the opposite
• Successful transcriptions of videos using Dirac notation, eigenstate derivations, and vector calculus indicate generalized capability beyond introductory physics — related
• Abandoning preamble coding macros in favor of amsmath, unicode-math, and siunitx enabled successive error-free transcriptions — related
• Three recurring AI failure modes arose in this project-based learning context: plausible-but-incorrect code, missing specialized knowledge, and limited long-term context — related