Refactoring smelly Java code and adding context, role, or token-limit prompts cut LLM chain-of-thought token use by roughly 15-50% in this study, but the 'no quality loss' claim rests only on shallow similarity metrics.
AI in Software Engineering: Perceived Roles and Their Impact on Adoption
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
This paper investigates how developers conceptualize AI-powered Development Tools and how these role attributions influence technology acceptance. Through qualitative analysis of 38 interviews and a quantitative survey with 102 participants, we identify two primary Mental Models: AI as an inanimate tool and AI as a human-like teammate. Factor analysis further groups AI roles into Support Roles (e.g., assistant, reference guide) and Expert Roles (e.g., advisor, problem solver). We find that assigning multiple roles to AI correlates positively with Perceived Usefulness and Perceived Ease of Use, indicating that diverse conceptualizations enhance AI adoption. These insights suggest that AI4SE tools should accommodate varying user expectations through adaptive design strategies that align with different Mental Models.
fields
cs.SE 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Optimizing Token Consumption in LLMs: A Nano Surge Approach for Code Reasoning Efficiency
Refactoring smelly Java code and adding context, role, or token-limit prompts cut LLM chain-of-thought token use by roughly 15-50% in this study, but the 'no quality loss' claim rests only on shallow similarity metrics.