Paper Type

ERF

Abstract

Artificial intelligence (AI) development platforms let non-programmers build software by describing it in natural language. Three decades of conceptual modeling research show that specification format affects what gets communicated to human developers. Whether format effects transfer when the developer is an AI system, with a different processing architecture, is open. Ten undergraduate teams (N=50) built applications on Base44 using plain text, then rebuilt them using user stories with acceptance criteria or visual models. User stories outperformed visual models (Cohen's d = 0.49) and shifted errors from logic failures to layout problems. Specification quality predicted first-attempt success (Spearman r = 0.70), but students who articulated effective specification principles did not produce them. The findings identify the developer's processing architecture as a variable the conceptual modeling tradition needs to account for.

Paper Number

1923

Comments

AI SYSTEM

Download

Share

COinS
 
Aug 15th, 12:00 AM

Specification Format Effects When the Developer Changes from Human to AI

Artificial intelligence (AI) development platforms let non-programmers build software by describing it in natural language. Three decades of conceptual modeling research show that specification format affects what gets communicated to human developers. Whether format effects transfer when the developer is an AI system, with a different processing architecture, is open. Ten undergraduate teams (N=50) built applications on Base44 using plain text, then rebuilt them using user stories with acceptance criteria or visual models. User stories outperformed visual models (Cohen's d = 0.49) and shifted errors from logic failures to layout problems. Specification quality predicted first-attempt success (Spearman r = 0.70), but students who articulated effective specification principles did not produce them. The findings identify the developer's processing architecture as a variable the conceptual modeling tradition needs to account for.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.