Predicting Human-GenAI Collaboration Effectiveness: A Machine Learning Investigation Of Skill Configurations, Trust, And Work Design

Abstract

Generative artificial intelligence is an even more significant technology in the workplace, but not every employee is effective in cooperation with such systems. This generates a practical and research problem since variations in the user ability, trust and work circumstances can influence whether human-GenAI interaction can enhance performance or can result in constrained value provision. The research will investigate which factors should be used to predict effectiveness of a human-GenAI collaboration in terms of skill configurations, trust in GenAI, and work design. Quantitative design was used with a positivist, deductive, cross-sectional design, and questionnaires were also used as methods of data collection by involving 118 respondents who had previously used the GenAI tools.

The data were analysed in [1]SPSS with descriptive statistics, Cronbach alpha, Pearson correlation, multiple regression, independent samples t-test, and one-way ANOVA. The four scales had acceptable to strong reliability, and Cronbach alpha of skill configurations, trust, work design, and collaboration effectiveness are.866, .816, and.790, respectively. The regression equation was significant and it accounted 33.5 percent of the variance in collaboration effectiveness (R 2 =.335, F = 19.124, p <.001). The most predictable was work design (=.343, p <.001), and then comes trust (=.317, p =.001), and the skill configurations (=.228, p =.005).

The research is valuable in demonstrating that human capability, trust and organizational work structure determine an effective human-GenAI collaboration and not technology usage.