The Pulse
IBM study finds few AI chiefs control investment decisions
Just 18% of surveyed chief AI officers and equivalent leaders say they fully control AI investment decisions, an IBM study found. The survey also found differences in reported outcomes between a group IBM calls high-impact leaders and other respondents.

Only 18% of surveyed chief AI officers and equivalent enterprise AI leaders say they have full control over AI investment decisions, according to an IBM Institute for Business Value study. The survey also found that many respondents do not see their organizations acting quickly when AI performance data points to opportunities for improvement.
Source newsroom.ibm.com
The institute conducted the survey with Oxford Economics, questioning 1,667 senior executives responsible for enterprise AI strategy, governance, implementation and value realization. It ran from April to June 2026 across 33 geographies and 20 industries; IBM released its findings on November 17.
Decisions and resources
The respondents included chief AI officers, chief data and AI officers, heads of AI and equivalent leaders. The study refers to the group collectively as chief AI officers, or CAIOs.
Just 29% of respondents said their organizations quickly adjust decisions and resources when AI performance insights reveal opportunities to improve outcomes, the study found. Another 40% said decisions are discussed in those circumstances but action is delayed.
Only 31% of all surveyed CAIOs said clear thresholds for stopping AI initiatives are critical to AI success. In the study’s foreword, Joanne Wright, IBM’s senior vice president for transformation and operations, wrote that AI leaders’ outcomes often depend on decisions made by others. She wrote that successful leadership requires “gathering evidence, influencing decisions, and measuring success.”
Autonomy and oversight
Some 48% of respondents said AI agents initiate and execute actions across workflows, according to the study. At the same time, 75% said governance, risk or compliance delays AI-driven action.
The survey found that 73% had defined where human oversight is non-negotiable. But 61% said AI decision rights were clearly defined and consistently applied.
Leaders in the group IBM calls high-impact CAIOs were nearly four times as likely to say they could quickly pause, adjust or redirect AI systems without disrupting operations. Among that group, 96% said AI-generated outputs could trigger actions autonomously, compared with 40% of all other CAIOs.
Business models and workflows
Thirty-two percent of surveyed leaders said AI was reshaping their business model, while 35% said it primarily strengthened the existing model. Another 26% expected AI to reshape the model but said their organizations had not made the necessary shifts.
IBM said 71% of CAIOs believed their organizations risked falling behind competitors in AI transformation. Respondents ranked workflows and business processes as their top priority for change.
Nearly all high-impact CAIOs said AI was embedded directly into workflows, compared with 59% of all other respondents. The study also examined organizational behaviors associated with stronger enterprise outcomes, including workflow integration, decision rights and business-model innovation.
Results reported by high-impact leaders
That group reported that 63% of AI projects delivered measurable business outcomes, compared with 43% among all other CAIOs. They also reported that 49% of projects moved from pilot to production, versus 33% for the others. Successful initiatives took eight months to scale enterprise-wide, compared with 13.5 months among the other respondents.
Among high-impact leaders, 94% said they used business outcomes to determine which initiatives received additional funding and talent. Some 96% of that group said they stopped or reduced investment in initiatives that failed to deliver, even when those initiatives had senior sponsorship. The equivalent figure among other respondents was 46%.
The study links the stronger reported outcomes with differences in how organizations manage AI autonomy and control, redesign workflows and make investment decisions based on evidence.
Source
Companies in this story
Related stories
- IBM survey finds gap between confidence and proof of digital control
- Stanford finds job concerns persist as AI investment tops $1 trillion
- IBM and Microsoft announce Imagine Enterprise agent platform
- SoftBank seeks over $11 billion in junk bonds for OpenAI bet
- IBM releases Granite 4.1 3B LLMON model in preview