Leadership

Why More AI Investment Won’t Fix Poor Business Decisions

Companies are committing resources to artificial intelligence, anticipating opportunities to automate work, accelerate reporting, and improve productivity. McKinsey’s 2026 Global Tech Agenda identifies AI as the leading technology investment priority

Adam Bent
Nov 19, 2026 · 10:08 AM · 6 min read
Share
Why More AI Investment Won’t Fix Poor Business Decisions

Companies are committing resources to artificial intelligence, anticipating opportunities to automate work, accelerate reporting, and improve productivity. McKinsey’s 2026 Global Tech Agenda identifies AI as the leading technology investment priority for the next two years, with half of surveyed companies naming it a top investment area. 

Yet investment and operational integration represent different stages of organizational change. OneTrust’s 2026 AI-Ready Governance Survey found that 74% reported departmental or scaled AI adoption, while only 5% said coordination and accountability were clear across the AI lifecycle. These findings raise a leadership question of how organizations can translate AI investment into better decisions and enterprise outcomes.

Danilo Arba, an AI transformation practitioner, researcher, and author, believes the answer begins with examining how organizations make decisions. With more than 25 years of experience managing major international programs, including portfolios exceeding $100 billion, Arba has built or transformed over 50 project management offices (PMOs). He sees automation as a source of immediate gains, while deeper transformation depends on how work and decisions are organized.

“If you simply add AI to the way your organization already works, you may just end up doing the same things faster,” Arba says. “The real opportunity is to rethink how decisions are made, who is responsible for them, and how they translate into actual business results.”

This perspective has implications for traditional operating models. Many PMOs have developed around reporting hierarchies, administrative coordination, compliance, and sequential approvals. Their established responsibilities may, however, leave limited capacity for interpreting emerging signals, anticipating consequences, and convening decision-makers early enough to influence outcomes.

Introducing AI tools can improve information processing, although the benefits depend on the organizational arrangements surrounding them. Fragmented datasets, overlapping responsibilities, and unclear decision rights can affect how machine-generated insights translate into action. Arba therefore encourages executives to reconsider the conventional emphasis on PMO maturity. Procedures can indicate development, while practical value depends on a PMO’s contribution to enterprise performance. A useful question, he suggests, is how decisions and outcomes improve through its involvement.

His experience assessing PMOs informs a framework describing three possible directions. Administrative PMOs collect information, coordinate reports, and maintain standard processes. Assurance-oriented PMOs emphasize oversight, compliance, and risk management. A third possibility, the decision-intelligence PMO, interprets information across programs, anticipates emerging challenges, and helps leaders determine where intervention may generate value. The third model offers an opportunity to connect project activity with enterprise priorities.

This evolution changes how organizations use project information. Reports, schedules, financial records, correspondence, site observations, and risk registers may contain signals of emerging patterns. When connected through shared data structures and identifiable ownership, these sources can help AI systems retrieve information, identify relationships, compare developments, and prepare scenarios. The PMO can then support a forward-looking decision process, helping executives examine potential consequences while intervention remains possible.

Arba draws on his cross-sector experience to illustrate how AI could improve decision-making. In an AI program currently under development, he is exploring how information from different project sources can be connected to identify emerging problems earlier. A potential delay might first appear in a site voice note, be documented differently in a subsequent report, and only become linked to its financial implications weeks later. By connecting these signals through a shared data structure, AI could help project teams recognize developing risks sooner, assess their potential consequences, and intervene before problems escalate.

“AI can connect information that would otherwise remain scattered across an organization. The value lies in turning those signals into useful insights, better decisions, and knowledge that informs future projects,” Arba explains.

These principles also inform Arba’s book, Thinking with Machines: Building AI-Enabled PMOs for the Age of Intelligent Project Leadership. Drawing on his program leadership and doctoral research, the book translates decision-making theory into practical operating frameworks. It examines how decision rights, evidence, governance, and human-machine responsibilities can be integrated into everyday business execution. Moreover, it offers guidance for developing PMOs that contribute to enterprise intelligence and strategic outcomes.

Putting this thinking into practice begins with examining value. Arba proposes that leaders identify where their organization generates value and where resources or opportunities may be lost. They can then pinpoint the decisions influencing those outcomes, determine the capabilities and responsibilities required, examine how information and authority move through existing processes, and establish the necessary data foundations. Technology selection follows once these elements have been considered.

The sequence also helps determine which responsibilities can be assigned to AI. Systems may be suited to monitoring information, identifying patterns, preparing forecasts, and generating recommendations. Human leaders retain an essential role in interpreting context, evaluating priorities, and accepting responsibility for consequential decisions. Arba advocates explicit boundaries through a framework defining an AI agent’s identity, permitted responsibilities, operating thresholds, and accountable human owner.

The degree of autonomy can reflect a decision’s potential consequences and reversibility. Even decisions with potentially significant consequences may be suitable for carefully governed AI execution when their effects can be readily reversed and appropriate safeguards are in place. Decisions involving substantial financial exposure, lasting operational consequences, or limited scope for reversal warrant human authorization.

For leaders considering where to begin, Arba suggests selecting a business decision with meaningful implications for organizational value. Mapping its supporting information, ownership, authority, and human-machine responsibilities can provide a practical starting point for testing a redesigned decision process before extending it elsewhere.

As intelligent systems become more integrated into organizational work, Arba believes an important part of transformation involves helping employees understand how AI can enhance their capabilities rather than viewing it primarily as a threat to their jobs. While AI may take on more responsibility for processing information and preparing actions, professionals can devote greater attention to interpretation, strategic judgment, and leadership. Achieving this shift requires more than introducing new technology. Organizations must invest in developing employees’ understanding of AI, addressing concerns about displacement, and helping people recognize opportunities to improve how they work.

“Technology can extend the intelligence available to an organization, while leadership determines how that intelligence is applied. The opportunity is to build institutions where people and machines contribute to better decisions, with responsibility remaining clear at every stage,” he remarks.

For organizations navigating the next phase of AI adoption, this perspective invites a deeper examination of investment, organizational design, and enterprise value. AI may deliver broader benefits as leaders reconsider the decisions governing their operations, equip people to work alongside intelligent systems, and establish structures through which insights become accountable action.

Adam Bent
Written by
Adam Bent is a business and technology writer with a focus on innovation, entrepreneurship, and sustainable business practices. His articles examine how emerging ideas, strategic leadership, and evolving industries are driving long-term growth and impact.
CEO Monthly is verified by MuckRack's editorial team. learn more
01
Newest first

Continue reading

Business
·
Business
·
Leadership
·
Blog
·