In the age of AI, competitive advantage flows from the integration of people and technology – and leadership is decisive
Writing Tsihilidzi Marwala & Sharmla Chetty
For more than a century, economists, executives and policymakers have explained organizational success through the concept of capital. Industrial firms relied on physical capital. The knowledge economy elevated human capital. Today, however, the rise of artificial intelligence (AI), machine learning and autonomous systems is forcing us to reconsider the foundations of value creation.
We are entering an era in which organizational performance depends not only on people, but on the interaction between human intelligence, artificial intelligence and autonomous agents operating under effective leadership. This emerging reality calls for a new framework: Performance Intelligence.¹
The industrial age taught organizations to optimize labor, machinery and processes. Human capital became the dominant lens through which talent was understood and managed. Recruitment, training and performance measurement were designed to maximize the productivity of employees. Yet the assumptions underlying this framework are increasingly inadequate. In many organizations, critical decisions are informed by AI systems, routine activities are automated, and autonomous agents execute tasks with limited human intervention. The workforce is no longer exclusively human.
The AI challenge today is no longer adoption, but value creation. McKinsey’s 2025 report, The State of AI, found that while 88% of organizations have deployed AI in at least one function, only 39% report any measurable impact on earnings. BCG’s 2025 global study, The Widening AI Value Gap, paints an even sharper picture: 60% of companies generate little or no material value from their AI investments. The gap is even starker at the top. Only 5% – just one in 20 – qualify as what BCG calls “future-built.” These firms achieve 1.7 times the revenue growth and 3.6 times the three-year total shareholder return of laggards, demonstrating that AI leadership translates into superior business performance. Meanwhile, Deloitte’s 2026 Global Human Capital Trends study, drawing on more than 9,000 leaders across 89 countries, identifies the root cause: organizations taking a technology-first approach are 1.6 times more likely to miss their AI return targets than those taking a human-centric approach. These failures are not technological. They are architectural.
To understand the transition, it is useful to revisit the work of Herbert Simon, whose theory of bounded rationality transformed economics and decision science. Simon argued that humans are constrained by limited information, finite cognitive capacity and imperfect judgment. Rather than optimizing, people satisfice: they settle for ‘good enough.’ This insight fundamentally changed how scholars understood decision-making.
Advances in artificial intelligence create an opportunity to rethink these constraints. In Artificial Intelligence and Economic Theory (Marwala and Hurwitz, 2017), the concept of flexibly-bounded rationality was introduced to explain how technology can expand the limits of human decision-making. As data availability increases, computational power grows and algorithms improve, some of the boundaries that restrict rationality become more flexible. Humans remain bounded, but technology extends the scope and quality of decisions.
The Performance Intelligence equation
This shift has profound implications for organizations. If machines can augment cognition, identify patterns at scale and process information consistently, then performance can no longer be understood solely through the capabilities of individuals. Instead, performance emerges from an integrated system of intelligence. This leads to a new formulation for Performance Intelligence.
PI = (w1HI + w2SI + w3AI + w4AA) × L
Performance Intelligence (PI) represents the integrated output of an organization. Human Intelligence (HI) contributes creativity, empathy, ethics, contextual judgment and strategic imagination. Social Intelligence (SI) encompasses psychological safety, interpersonal trust and team cohesion – the collective conditions that determine whether human and machine capabilities can be effectively combined. AI contributes pattern recognition, forecasting, optimization and analytical consistency. Autonomous Agents (AA) contribute execution, adaptation and operational scalability. Leadership (L) acts as the multiplier that aligns these capabilities toward a shared purpose.
The structure of the equation is important. Human intelligence, social intelligence, artificial intelligence and autonomous agents are additive: each contributes a distinct capability. Leadership is multiplicative, because it determines whether these capabilities reinforce one another or remain disconnected assets. Without effective leadership, organizations risk deploying powerful technologies without direction, coordination or trust.
The weights in the formula, which sum to 1, refine the calculation. They should be calibrated by function and sector to reflect the varying needs of different industries. In high-contact sectors, such as healthcare, professional services or education, Human Intelligence remains the primary performance driver. In data-intensive functions, such as algorithmic trading, logistics optimization or fraud detection, AI delivers the greatest marginal return.
Social Intelligence warrants particular attention from leaders. Amy Edmondson’s research, including her 1999 paper ‘Psychological safety and learning behavior in work teams’ – replicated across three decades and multiple sectors – identifies psychological safety as the strongest single predictor of team performance in complex and uncertain environments. Google’s Project Aristotle, a study of 180 internal teams, reached the same conclusion. Yet Deloitte’s 2026 research found that 34% of organizations name culture as the primary barrier to AI transformation. Culture is not a soft supplement to organizational performance in the AI era. It is a structural input.
Intelligence, logic and the leadership multiplier
The concept of Performance Intelligence challenges traditional assumptions about productivity. Historically, productivity measured the relationship between inputs and outputs. The objective was to produce more with less. In the intelligence age, however, the challenge is not simply efficiency. It is the effective combination of multiple forms of intelligence.
Organizations increasingly face questions that cannot be answered by human expertise alone. Climate modeling, financial forecasting, healthcare diagnostics and supply chain optimization all require computational capabilities beyond the reach of individuals. At the same time, organizations cannot rely exclusively on machines. Ethical dilemmas, strategic trade-offs and questions of legitimacy require human judgment. This creates a new managerial challenge: determining how responsibilities should be allocated between humans, AI systems and autonomous agents. The most successful organizations will not be those that maximize automation, but those that optimize collaboration.
A critical distinction must be made between logic and rationality. Machines can be highly logical: they process data according to defined rules and objective functions. Yet they are not necessarily rational in the broader sense. Models are simplifications of reality. Data may be incomplete. Objectives are specified by humans. Context is often difficult to encode. As a result, there remains a gap between logical consistency and true rationality.
The leadership multiplier
The gap between logic and rationality underlines the enduring importance of leadership – the multiplier that makes every other investment compound or collapse.
Leaders provide purpose, values, and accountability. They ask whether an optimization objective is appropriate. They determine which risks are acceptable. They decide when efficiency should yield to fairness, resilience or social responsibility. Leadership becomes more important, not less, as AI becomes more capable.
To lead effectively in the PI era requires three specific competencies. First, AI literacy: the ability to evaluate, direct and interrogate AI outputs without becoming a technical specialist, including understanding model limitations, data biases and failure modes. The second is orchestration: the capacity to design and govern workflows in which human judgment, AI analysis and autonomous execution are correctly sequenced, with clear decision rights at each stage. Third, leaders need to understand and develop trust architecture, building psychological safety, transparency and accountability structures that allow human-AI teams to perform under uncertainty.
This points to a structural failure of accountability running beneath the adoption gap. Deloitte’s 2026 study found that 56% of leaders design AI systems solely for business outcomes, while only 40% design for both business and human outcomes. 42% of workers report that their organizations do not evaluate AI’s impact on people at all. The cumulative effect is described as “cultural debt,” the performance liability that accumulates when organizations optimize for AI while neglecting the people systems around it. The leadership multiplier can only function positively when accountability for human outcomes is built into the governance structure alongside accountability for business outcomes.
Organizations that neglect this are not merely failing ethically; they are accumulating a risk that will eventually constrain performance regardless of AI investment.
The rise of autonomous agents
The emergence of autonomous agents introduces additional challenges. Unlike traditional software, autonomous agents can act, adapt and learn, and they are already a significant source of organizational value. BCG’s 2025 study found that agentic AI accounts for 17% of total AI value creation today, a share expected to reach 29% by 2028. Future-built companies are three times as likely to deploy agents as laggards. In Seizing the Agentic AI Advantage (2025) McKinsey identifies agentic AI as the next frontier for competitive differentiation, with the most advanced organizations already moving from single-function agents to multi-agent systems capable of complex, multi-step reasoning.
Their growing role requires governance built in advance, not retrofitted after problems emerge. Organizations must establish clear boundaries regarding what agents can do, what decisions require human approval, and how responsibility is allocated when outcomes are unexpected. Trust in AI systems must be earned through transparency, validation, monitoring and accountability: it cannot be assumed.
Talent and intelligence literacy
The rise of Performance Intelligence also changes how organizations must think about talent. Traditional approaches focus on individual competencies; future-oriented organizations must develop collective intelligence. BCG’s analysis found that companies realizing the most value from AI have the most targeted upskilling programs, and points out that 70% of AI’s total potential value is concentrated in core business functions including R&D, product innovation and digital marketing.
Three tiers of intelligence literacy are required. At the senior leadership level, strategic literacy means the ability to make AI investment decisions, set governance policy and hold AI systems accountable for both business and human outcomes. At the management level, operational literacy means the ability to redesign workflows around human-AI collaboration and interpret algorithmic outputs in context. At the front line, task literacy means the practical skills to direct, interrogate and override AI outputs in daily work. Performance management systems must evolve accordingly: evaluating employees without evaluating the AI systems they use provides an incomplete picture of organizational capability.
Competing through intelligence
The shift from human capital to Performance Intelligence does not diminish the value of people. On the contrary, it highlights uniquely human capabilities. Creativity, empathy, ethical reasoning and the ability to navigate ambiguity remain indispensable. The future belongs neither to humans nor to machines alone. It belongs to systems that combine the strengths of both.
The implications extend far beyond individual organizations. Investors may increasingly assess organizations by their intelligence architectures. Nations may compete on their ability to build ecosystems that support Performance Intelligence, including education systems, digital infrastructure, governance frameworks and innovation capacity. Universities must prepare graduates not only to acquire knowledge, but also to collaborate effectively with intelligent technologies. Intelligence literacy may become as important as digital literacy.
For generations, firms competed on scale. Later, they competed on information and knowledge. Today, they are beginning to compete on intelligence. The rise of Performance Intelligence signals a fundamental transformation in how value is created, measured and sustained. The organizations that thrive will be those that recognize that the quality of integration between people and machines is now the most important source of competitive advantage.
Professor Tshilidzi Marwala is Rector of the United Nations University and UN Under-Secretary-General, and co-author with Evan Hurwitz of Artificial Intelligence and Economic Theory
Dr Sharmla Chetty is CEO of Duke Corporate Education
REFERENCE
- Adelanwa, A, Basnet, A, and Anene, UN (2023) ‘Performance Intelligence Models for Optimization and Outcome Measurement in Large Scale Public Service’, https://shisrrj.com/paper/SHISRRJ23680.pdf
Image credit: Shutterstock
