Scaling AI successfully and responsibly: Insights from Maven Insights analysis
While artificial intelligence (AI) has shifted from being a nice-to-have to an imperative, leaders are now grappling with the challenge of scaling AI across their organizations. New research from Maven Insights examines how leaders can successfully and responsibly scale the technology in order to reap its true value.
The report, titled ‘Scaling Responsible AI Adoption’ and authored by Gagan Arora (Partner), Mohammed Sheet (Manager), Jack Ghazi (Senior Consultant), and Furkan Yıldırım (Consultant), examines global trends in AI planning and implementation and identifies what differentiates leading organizations from the rest.
Global adoption and the value gap
The survey confirms that AI has entered the mainstream. 88% of organizations now use AI in at least one business function, and 62% are experimenting with AI agents.
Progress is expected to continue unabated. Gartner forecasts that worldwide AI investment will exceed $2.5 trillion by 2026, with generative AI as a major driver of growth. Spending on generative AI alone reached approximately $644 billion in 2025, up 76% from the previous year.
Despite this rapid adoption, the business impact of AI remains limited. Maven Insights references McKinsey data showing that only 39% of organizations report any improvement in earnings, and nearly two-thirds have not scaled AI across the enterprise.
A study by US-based researcher MIT paints an even more striking picture: 95% of organizations see no measurable profit and loss impact from AI, and only about 5% of AI pilots extract meaningful value. The same study notes that while more than 80% of organizations have experimented with ChatGPT or Copilot tools, custom enterprise AI systems rarely progress beyond the lab, with only 5% reaching production.
One of the key reasons MIT highlights for failed AI rollouts is that leaders often allocate budgets toward flashy, low-value use cases rather than the highest-impact initiatives. Other common challenges include lack of clear ownership, unclear governance frameworks, and pilot projects that fail due to inadequate project or technical delivery. In addition, systems or algorithms may be deployed, but end-users often do not adopt them properly, further limiting impact.
In light of these obstacles, Gartner predicts that by 2025, 30% of generative AI projects will be abandoned after proof-of-concept.
These developments underscore the “value gap” between AI ambition and execution – a trend that Maven Insights says it also observes firsthand in its consulting engagements.
Why value remains limited
The authors identified several bottlenecks in AI’s return on investment, and notably, most of these are not technical or algorithm-focused but instead organizational topics. Five obstacles that stand out are:
- Data and trust breakdowns
- Capability and literacy gaps
- Regulatory and governance complexity
- Operating models built for stability, not learning
- Cultural resistance and shadow AI
Building on these and other insights, Maven Insights proposes a framework for successful AI adoption. “Organizations that outperform in AI transformation take a different path: they invest first in trusted data and transparent governance, build broad AI literacy across managers and staff, and embed governance into every stage of execution.”
“They treat AI as an enterprise capability tied to clear business metrics, not a collection of isolated experiments.”
A case study in Saudi Arabia
The report highlights Saudi Arabia as a best-practice example of coordinated AI adoption at a national level. “Under Vision 2030, 66 of the 96 national objectives relate to data and AI. The Kingdom hosts 33 existing data centers and 42 under development, with upcoming capacity of approximately 2.2 gigawatts.”
This infrastructure ambition is matched by talent development. Eighty-six percent of Saudi universities now offer AI undergraduate degrees, and more than 45,000 professionals have been trained through SDAIA programs. The national curriculum integrates data and AI literacy across disciplines, and the strategy targets 20,000 AI and data specialists by 2030.
The development drive is also diversity-focused. The Elevate initiative, for example, aims to train over 25,000 women in data and AI.
“This unified approach, coupled with abundant energy for power‑hungry data centers, positions Saudi Arabia as a leader in the AI landscape and enables the Kingdom to convert AI adoption into economic value for businesses and society.”
From AI adoption to AI accountability
The Maven Insights analysis concludes that AI has evolved from an experimental discipline to a professional management discipline. Organizations that thrive are those that assign clear ownership, embed governance and ethics into delivery, invest in data quality and workforce skills, and rigorously measure outcomes.
“With these foundations in place, leaders can ensure that AI investments deliver the value they promise, generating benefits across their organizations.”
