Publication Type

PhD Dissertation

Version

publishedVersion

Publication Date

4-2026

Abstract

As a novel technological instrument for content production and customer communication, generative artificial intelligence (GenAI) has garnered significant attention for its value in private domain marketing. However, enterprises frequentlyencounter a paradox of high initial introduction rates coupled with a lackof substantive application. Employee apprehensions regarding the qualityof AI-generated content (AIGC) and organizational guidance strategies are critical factors impeding the realization of technological value. Drawing upon the technologyacceptance model (TAM), the unified theory of acceptance and use of technology(UTAUT), and the task-technology fit (TTF), the study constructs a theoretical model positioning corporate AI guidance strategies as the independent variable andemployee AI acceptance as the dependent variable. The model incorporates theperception of GenAI-generated content (accuracy, credibility, and business fit) andGenAI system objective accuracy as moderating variables. Employee AI acceptanceis operationalized through three progressive dimensions: usage rate, adoption rate, andcustomer conversion rate. Usage rate reflects the behavioral frequencyof employee-AI interaction. Adoption rate manifests the extent of value recognitionandtrust in AI outputs. Customer conversion rate denotes the performance efficacyof AI-enabled business objectives. These three dimensions follow a sequential logicwherein usage rate establishes the foundation, adoption rate facilitates conversion, andconversion rate supports performance realization. The study investigates thedifferential effects of mandatory usage strategies on these dimensions, examines theboundary conditions of subjective quality perceptions versus objective output performance, and identifies the mediating mechanisms of usage intensity and adoptionbehavior within the intervention process.

During the research process, a systematic literature review on technology acceptanceand human-AI collaboration is conducted to ground the research hypotheses, followedby the development of a theoretical model to clarify the logical relationships amongvariables. A field experiment is performed involving employees in private domainmarketing roles at company SFRH. Using a randomized controlled trial (RCT) design, 127 eligible employees are assigned to experimental and control groups, yielding122valid samples. The data sources include system logs, text comparisons, business performance records, expert ratings, and questionnaire surveys. Empirical analysis onthe data is conducted using regression analysis, quantile regression, moderation effect testing, and mediation effect testing. Upon hypothesis verification, the followingconclusions are drawn: First, mandatory usage strategies significantly boost AI usageinvestment, adoption, and conversion. However, this effect displays a temporal decay, with strong initial momentum followed by diminishing marginal returns. Furthermore, the impact is primarily concentrated among employees at moderate usage levels, withlimited marginal efficacy for those at the low and high quantiles. Second, subjectivequality perception of GenAI-generated content positively moderates the impact of mandatory intervention on usage investment and customer conversion rate, yet its moderating effect on adoption is insignificant. Objective accuracy similarly enhances these impacts, revealing a quality threshold effect: Mandatory strategies yieldnet positive performance gains only when employee evaluations of output quality exceeda moderate threshold. Third, usage intensity is the primary mediator, explainingapproximately 60% of the indirect effect on customer conversion rate, whereas themediating role of adoption rate is notably weaker.

Based on these findings, the study offers three managerial implications: First, mandatory usage should be institutionalized as a persistent operational protocol rather than a one-time policy. This involves default system activation settings, incorporating intensity metrics into operational dashboards, and mandatingAI justification for resource approval, ensuring the implementation of employee usageinvestment. Second, quality thresholds should be integrated with mandatorypromotion. Minimum quality standards should be established for critical business scenarios, retaining manual review for sub-standard outputs to prevent workloadinflation and resistance. Third, enhancing adoption rates requires minimizingprocedural friction. Transforming AI suggestions into structured, reusable script components, along with the compliance verification mechanisms, moves adoptionfrom discretionary judgment to system-guaranteed standardized execution, therebybridging the gap between usage investment and business performance.

Keywords

generative artificial intelligence (GenAI), intervention strategies, employee AI acceptance, GenAI content quality

Degree Awarded

Doctor of Business Administration (Accounting and Finance)

Discipline

Marketing | Technology and Innovation

Supervisor(s)

LI, Linyi; YUE, Heng

First Page

1

Last Page

236

Publisher

Singapore Management University

City or Country

Singapore

Copyright Owner and License

Author

Download

Share

COinS