About Beyond the Hype: Navigating the Real-World Pipeline of Enterprise AI Adoption in Journal of Open Innovation: Technology, Market, and Complexity.
For years, the conversation around Artificial Intelligence (AI) has focused on a simple question: "Are you using it or not?" However, a new study of 816 Slovak enterprises reveals that AI diffusion is far more complex than a simple binary switch. By viewing AI adoption through the lens of open innovation dynamics, researchers have developed a "Staged Markov Model" to forecast how firms actually move from awareness to implementation.
The Three States of AI Readiness
The research moves away from traditional "all-or-nothing" models, instead classifying enterprises into three distinct states:
- Non-adopters: Firms with no current AI and no active plans to adopt.
- Planners: The "missing middle"—firms actively preparing or piloting AI solutions within a three-year horizon.
- Adopters: Firms that have fully integrated AI into their core or selected processes
The Two Critical Engines: Activation and Conversion
The study identifies that AI spreads through two primary mechanisms. First is activation, where non-adopters are persuaded to start planning (N→P). Second is conversion, where planners successfully transition into full adoption (P→A).
This distinction is vital for managers and policymakers because the barriers are different for each. While "activation" might require better awareness, "conversion" is often the real bottleneck, requiring deep organizational changes, new skills, and robust data governance.
The De-adoption Reality Check
Perhaps the most striking finding is that AI adoption is not always permanent. While many models assume that once a firm adopts AI, they stay adopters forever ("absorbing adoption"), this study introduces a de-adoption variant.
Drawing on European economic data, the model accounts for a 5% annual de-adoption rate, where firms stop using AI due to high costs, low ROI, or governance friction. When de-adoption is factored in, the market never reaches 100% saturation. Instead, it hits a stationary equilibrium—a ceiling where new adopters are balanced out by those abandoning the technology.
Key Findings for Leaders and Policymakers
- AI adoption is bounded, Not Self-Sustaining: In scenarios where firms can "de-adopt," the long-run adoption level stabilizes between 72% and 88% depending on the economic environment, rather than reaching the whole market.
- The "Pilot Trap": Many firms remain in the "planner" state for extended periods. Diffusion speed depends heavily on how quickly these firms can move beyond experimentation into routinized use.
- Retention is as important as Adoption: Results show that reducing the de-adoption rate (keeping firms using AI) can have a larger impact on long-term digital transformation than simply pushing new firms to start pilots.
- Scenario-Dependent Speed: Under "Optimistic" conditions, adoption shares can jump from 68% to over 80% in just one year, but "Conservative" environments see much more sluggish growth with a persistent pool of non-adopters.
Authors: Vladimír Hojdik, Peter Štetka, Nora Grisáková, Zuzana Hajduová, Federico Banda, Diana Pallérová (2026). AI diffusion as open innovation dynamics: Forecasting adoption, de-adoption, and stationary equilibrium in slovak enterprises, IN Journal of Open Innovation: Technology, Market, and Complexity, Vol. 12(3). https://doi.org/10.1016/j.joitmc.2026.100839.
