ReadingResearchRadarInvestment framework
Sign in / Sign up
← Back to reading
NewsMaterial3 min readDisconfirmAI Editorial

IBM Study: Chief AI Officers See Where AI Creates Value, but Few Control Investment Decisions

An IBM Institute for Business Value survey of 1,667 corporate AI leaders reveals that only 18% of Chief AI Officers (CAIOs) have full control over AI investment decisions. High-impact CAIOs achieve significantly stronger AI outcomes by tying funding to business results, embedding AI directly into workflows, and retaining the ability to quickly intervene in AI systems.

2026-11-17 13:01 GMT+8

Conducted jointly by the IBM Institute for Business Value and Oxford Economics, this survey covers 33 regions and 20 industries. It surveyed 1,667 senior executives responsible for enterprise AI strategy, governance, implementation, and value realization, including Chief AI Officers (CAIOs), Chief Data and AI Officers (CDAIOs), Heads of AI, and equivalent roles. The study's core finding points to an organizational design issue rather than a technical one: these leaders can see where AI creates value and where it fails, but most lack the authority to reallocate resources based on those insights.

The specific numbers are stark: only 18% of respondents say they have full control over AI investment decisions; only 29% say their organizations can quickly adjust decisions and resources when AI performance insights reveal opportunities for improvement. This means more than 70% of CAIOs see the signals but cannot pull the levers. Joanne Wright, Senior Vice President of Transformation and Operations at IBM, writes in the report's foreword that few leaders have as broad a view of how AI creates value across the enterprise as CAIOs do, yet the outcomes they are held accountable for often depend on decisions made by others.

The study identifies a group of high-impact CAIOs whose organizations report significantly stronger results: 63% of their AI projects deliver measurable business outcomes, compared with 43% for other CAIOs; 49% move from pilot to production, versus 33%; and successful initiatives scale enterprise-wide in eight months, compared with 13.5 months for others. The gap between these three metrics is not a gap in technical capability, but a gap in decision-making mechanisms.

Sign in / Sign up
ReadingResearchRadarInvestment framework

High-impact CAIOs share three common, interconnected practices. First, they direct funding based on business results: 94% say they use business outcomes to determine which projects receive additional funding and talent, and 96% say they stop or reduce investment in initiatives that fail to deliver, even if they have senior sponsorship—compared with only 46% of other CAIOs. Second, they embed AI directly into workflows: nearly all high-impact CAIOs report AI is embedded in workflows, compared with 59% of others. Third, they retain the ability to quickly intervene in AI systems: high-impact CAIOs are nearly four times as likely to say they can quickly pause, adjust, or redirect AI systems without disrupting operations. At the same time, 96% allow AI outputs to autonomously trigger actions—a seeming contradiction that points to a single premise: expanding autonomy is only safe when clear decision rights and rapid intervention capabilities exist.

The boundaries of this study need to be clearly stated. All data comes from respondent self-reports, and the criteria for defining the high-impact group were set by the study itself; therefore, the causal direction—that these practices lead to better results—cannot be established from this cross-sectional survey. A equally plausible explanation is that organizations with higher overall management quality are both more willing to grant CAIOs investment decision rights and more likely to produce good AI projects. The study does not provide randomized controlled trials or longitudinal tracking to rule out this confounding factor. Additionally, the survey was conducted from April to June 2026, and respondents' judgments reflect the state of their organizations at that time, making it impossible to predict whether these practices will remain effective as AI capabilities expand in the future.

Sources

  • prnewswire.com↗
  • ibm.comIBM↗
Continue researching →