CA prototype technologies shifted from rule-based and deterministic systems toward open-source tools, pre-trained LLMs, and retrieval-augmented generation pipelines¶

Claim · All claims
Evidence · 1 study · review r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Across the reviewed studies, earlier implementations prioritized rule-based and deterministic systems while more recent work integrated open-source tools, pre-trained LLMs, and RAG pipelines for adaptive, context-sensitive interaction. → Barzanji and Loitsch 2025

Evidence¶

Barzanji and Loitsch 2025¶

Barzanji and Loitsch. (2025). Exploring conversational agents for novice programmers: a scoping review. Discover Artificial Intelligence. https://doi.org/10.1007/s44163-025-00521-4

q2 · i? · review · r2

The review's prototype technology analysis (RQ3), organized by a four-module architecture, reports that "earlier implementations pri -oritized rule-based and deterministic systems, while more recent work integrated open-source tools, pre-trained LLMs, and retrieval-augmented generation (RAG) pipelines". Table 3 clusters technologies into rule-based (n = 9), LLM-based (n = 8), and retrieval-augmented/hybrid (n = 2) paradigms.

"earlier implementations pri -oritized rule-based and deterministic systems, while more recent work integrated open-source tools, pre-trained LLMs, and retrieval-augmented generation (RAG) pipelines to support adaptive, context-sensitive interaction."

Discussion¶

Related Claims¶

• Most of the 23 reviewed CA studies for novice programmers relied on subjective usability measures, with only three using objective pre/post-test designs to measure knowledge gains — related
• The review identifies three pathways of AI-empowered STEM education: technological leap to generative AI, integration into instructional systems, and value shift toward competency cultivation — a broader claim this one bears on
• Respondents anticipated using LLMs most often (23%), followed by NLP (17%), predictive AI (14%), chatbots (13%), and ASR (10%) — related
• A hybrid human-AI workflow using GPT-4o with retrieval-augmented generation supported efficient inductive thematic analysis while preserving researcher judgment — 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
• Intelligent tutoring systems are a mature, technologically sophisticated implementation of AI-driven personalized learning with mostly positive outcomes in programming, mathematics, and STEM — related
• LLM capabilities enabled family education systems across conversational, embodied, and spatial interface forms to generate support from the context of an unfolding interaction — a narrower finding that bears on this claim
• SET-NLP shows expansion of the methodological repertoire without simple replacement of earlier approaches by newer ones — related
• Targeted enhancements (retrieval-augmented inference, multimodal training) substantially raise open-source models' phonology mastery to near-ceiling levels — a narrower finding that bears on this claim