Paper Number

1856

Paper Type

Short

Description

Research accepts that ML-based AI tools’ accuracy is a defining characteristic for AI implementation. Yet, the understanding of accuracy in relation to the “ground truth” remains under-researched, especially the understanding of universally recognized practices for the “ground truth” in specific knowledge domains. This short paper investigates how knowledge workers’ expertise can be used effectively to redefine the “ground truth” and produce training datasets conducive to more accurate ML predictions. It approaches the question empirically with a case study of ARUP, a global engineering and consultancy firm that uses various AI tools for its advisory services. The paper highlights how executives often overlook data preparation and the role of knowledge workers during this phase, thus questioning the meaning of “ground truth”. It provides valuable insights on how a total and constructive collaboration of stakeholders is essential for organizing existing data, contributing to existing literature on ML implementation and data in general.

Comments

10-AI

Share

COinS
 
Dec 12th, 12:00 AM

Machine Learning for ARUP: Time to Redefine the Ground Truth

Research accepts that ML-based AI tools’ accuracy is a defining characteristic for AI implementation. Yet, the understanding of accuracy in relation to the “ground truth” remains under-researched, especially the understanding of universally recognized practices for the “ground truth” in specific knowledge domains. This short paper investigates how knowledge workers’ expertise can be used effectively to redefine the “ground truth” and produce training datasets conducive to more accurate ML predictions. It approaches the question empirically with a case study of ARUP, a global engineering and consultancy firm that uses various AI tools for its advisory services. The paper highlights how executives often overlook data preparation and the role of knowledge workers during this phase, thus questioning the meaning of “ground truth”. It provides valuable insights on how a total and constructive collaboration of stakeholders is essential for organizing existing data, contributing to existing literature on ML implementation and data in general.

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