Paper Type
Complete
Abstract
This study presents an innovative methodology for literature identification and knowledge extraction using generative artificial intelligence (GAI) while addressing key challenges such as hallucinations, bias, transparency, and reproducibility. The analysis is focused on the Future of Work as shaped by GAI, a widely discussed topic with recurring themes including automation, augmentation, hallucination, bias, ethics, regulation, and dystopian futures. The proposed methodology systematically includes academic and industry publications to include contemporary perspectives while limiting selection bias. A training process for a custom GPT using natural-language reinforcement training for the identification and extraction of key topics is presented, providing a macro-level perspective from a large-scale literature review. The outlined process improves the harmonic mean of precision and recall by over 50% with three iterations, identifying the no-code process as effective. This approach applies to market research, financial analysis, political analysis, and public opinion analysis.
Paper Number
1993
Recommended Citation
Strohm, Charlie J.; Matta, Vic; and Bansal, Gaurav, "Increasing Generative AI Effectiveness through Triangulation: The Case for the Future of Work" (2025). AMCIS 2025 Proceedings. 22.
https://aisel.aisnet.org/amcis2025/data_science/sig_dsa/22
Increasing Generative AI Effectiveness through Triangulation: The Case for the Future of Work
This study presents an innovative methodology for literature identification and knowledge extraction using generative artificial intelligence (GAI) while addressing key challenges such as hallucinations, bias, transparency, and reproducibility. The analysis is focused on the Future of Work as shaped by GAI, a widely discussed topic with recurring themes including automation, augmentation, hallucination, bias, ethics, regulation, and dystopian futures. The proposed methodology systematically includes academic and industry publications to include contemporary perspectives while limiting selection bias. A training process for a custom GPT using natural-language reinforcement training for the identification and extraction of key topics is presented, providing a macro-level perspective from a large-scale literature review. The outlined process improves the harmonic mean of precision and recall by over 50% with three iterations, identifying the no-code process as effective. This approach applies to market research, financial analysis, political analysis, and public opinion analysis.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIGDSA