Firms adopt generative-AI CRM when they trust its functionality and task benefits, while information shortfalls and a preference for human contact slow uptake; neural-network analysis ranks perceived usefulness as the top predictor of trust and information barriers as the top predictor of inertia.

Main Finding

Organizations’ adoption of Generative AI–enabled CRM systems (GAICS) is driven by initial trust and inertia. Functional and instrumental values increase initial trust, while human interaction, information, and norm barriers increase organizational inertia. Both initial trust and inertia significantly affect GAICS adoption. Artificial neural network (ANN) analysis identifies: functional values as the strongest predictor of trust, information barriers as the strongest predictor of inertia, and the need for human interaction barriers as the strongest predictor of adoption.

Key Points

• Two-stage mixed-methods design:
  • Qualitative phase to identify critical factors shaping initial trust and inertia toward GAICS.
  • Quantitative phase to validate the conceptual framework.
• Positive drivers of initial trust: functional value (system usefulness/quality) and instrumental value (task-related benefits).
• Drivers of inertia (resistance to change): human interaction barriers (desire for human contact), information barriers (lack of knowledge/clarity), and norm barriers (cultural/social norms).
• Both initial trust and inertia have statistically significant effects on GAICS adoption decisions.
• ANN feature-importance results:
  • Functional values → most important predictor of initial trust.
  • Information barriers → most important predictor of inertia.
  • Need for human interaction barriers → most important predictor of adoption outcome.

Data & Methods

• Mixed-methods approach:
  • Qualitative exploratory work to surface determinants of trust and inertia in organizations adopting GAICS.
  • Quantitative validation (survey/empirical testing of the conceptual model).
• Analytical techniques:
  • Inferential analysis to test relationships (e.g., trust/inertia → adoption).
  • Artificial neural network (ANN) analysis used to rank predictors by relative importance for trust, inertia, and adoption.
• Notes:
  • Abstract does not report sample size, sectoral scope, or country context—these details are needed to assess external validity and generalizability.

Implications for AI Economics

• Diffusion and adoption dynamics:
  • Functional/instrumental value of AI systems speeds organizational adoption via increased trust — highlighting the economic importance of demonstrable productivity gains and clear ROI.
  • Information frictions (knowledge gaps) and human-interaction preferences slow adoption, acting as non-price coordination frictions that can delay diffusion and reduce realized aggregate productivity gains.
• Investment and cost-benefit considerations:
  • Firms should invest in user-focused functionality and clear value communication to boost trust and adoption.
  • Complementary investments (training, information campaigns, hybrid human–AI workflows) can reduce inertia and increase effective utilization of GAICS, improving returns.
• Labor market and organizational design:
  • Persistent human-interaction barriers imply tasks with high interpersonal content may resist full automation; firms will likely adopt hybrid models, affecting job reallocation and skill demand.
  • Flexible management practices that accommodate human–AI complementarities can lower switching costs and support smoother digital transformation.
• Policy and ecosystem:
  • Public support for information provision, standards, and best-practice guidance can reduce information barriers and accelerate socially beneficial adoption.
  • Regulators and industry bodies should consider norms and trust-building measures (transparency, accountability standards) to address cultural and norm barriers.
• Research priorities:
  • Quantify macroeconomic effects of inertia-driven delays in GAICS diffusion.
  • Estimate heterogeneity by industry/task type to model where GAICS generate net gains versus where human interaction frictions persist.