What Is the Human-in-Complexity Framework?

The Human-in-Complexity Framework™ is a leadership framework developed for environments where complexity, uncertainty, technology, and human judgment increasingly intersect. It is organized around five interconnected practices: Awareness, Judgment, Responsibility, Learning Forward, and Community.

Why does leadership need another framework?

The question is fair. Leadership is not short on models. Transformational leadership, servant leadership, adaptive leadership, sensemaking, and decades of decision research already give leaders useful ways to understand organizations and human behavior.

Human-in-Complexity is not intended to replace that work. It addresses a more specific problem that has become harder to ignore: leaders are increasingly operating inside systems where technology is not simply a tool in the background. It is becoming an active participant in how information is gathered, alternatives are generated, work is executed, and decisions are made.

That changes the operating environment, but it does not eliminate the fundamental leadership problem. Someone still has to determine what matters, interpret context, decide what risk is acceptable, reconcile competing priorities, and own what happens next.

This distinction builds on established work showing that leadership responses must change with the decision context and that complex settings differ from merely complicated ones (Snowden & Boone, 2007).

The framework grew from years of leading in operational environments where information was imperfect, conditions changed quickly, and waiting for certainty was rarely realistic. Academic study of organizational leadership, complexity, and decision-making gave me better language for patterns I had already experienced in practice. AI has made the underlying question more urgent: as the systems around leaders become more capable, what must remain fundamentally human about leadership?

The five practices

1. Awareness

Good judgment begins before the decision. Awareness is the discipline of developing the most accurate picture of the environment that the situation allows. It means distinguishing evidence from assumption, noticing changes in conditions, understanding the people and systems involved, and deliberately looking for information that contradicts the current view.

Technology can improve awareness dramatically. It can process more information than a person can, detect patterns, monitor conditions continuously, and retrieve relevant knowledge almost instantly. But more information is not the same as understanding. Leaders still have to determine what is relevant, what may be missing, and what the information means in context.

2. Judgment

Judgment matters most where the answer cannot simply be calculated. Leaders routinely face incomplete information, competing legitimate priorities, uncertain consequences, and decisions that affect people differently.

Experience matters here, but experience should be an input into judgment, not a substitute for it. What worked before can sharpen pattern recognition, but it can also create false confidence when the environment has changed.

AI can expand the decision space. It can surface alternatives, identify relationships, test scenarios, and expose information a leader might otherwise miss. Human judgment determines what should be done within that expanded space.

Research on human-AI decision making similarly describes complementary strengths: AI can extend analytical information processing while humans contribute capabilities needed when uncertainty and equivocality remain (Jarrahi, 2018). The broader management literature also cautions against treating automation and human augmentation as cleanly separable choices (Raisch & Krakowski, 2021).

3. Responsibility

Responsibility follows leadership. Organizations can delegate authority and automate work, but they cannot automate away accountability.

A machine-generated recommendation does not own the consequences of the decision. A workflow does not decide the values embedded in its design. Technology can participate in a decision, but it cannot absorb the leader’s responsibility for the system in which that decision was made.

This becomes more important, not less, as AI systems gain autonomy. Leaders must understand where authority has been delegated, what constraints surround that authority, how exceptions are handled, and who remains responsible when the system produces an outcome nobody intended.

4. Learning Forward

Leadership in complexity guarantees that some decisions will be wrong. The useful question is what the organization learns from them.

Learning Forward means converting experience into better future judgment. It requires separating the objective from the strategy, the strategy from its implementation, and the implementation from the outcome. A disappointing result does not automatically prove that the objective was wrong.

Never abandon a sound objective because of a flawed implementation. But never defend a flawed objective simply because you are invested in it.

The distinction matters because organizations often respond to failure by changing the wrong thing. Learning requires diagnosis before reaction.

5. Community

Organizations do not operate through processes alone. They operate through people. Trust, credibility, shared purpose, relationships, and candor affect how quickly information moves, whether problems are surfaced, whether people challenge weak assumptions, and whether decisions are actually carried out.

Community is therefore not a soft addition to the framework. It is part of the operating context that makes the other four practices possible. Awareness degrades when people hide information. Judgment degrades when disagreement is unsafe. Responsibility degrades when accountability becomes blame. Learning degrades when failure cannot be examined openly.

Edmondson’s (1999) research on psychological safety and team learning provides an important empirical foundation for this relational dimension, particularly the connection between interpersonal risk, learning behavior, and team performance.

How should leaders use AI inside the framework?

The purpose is not to protect human work. If technology can perform a task better, faster, safer, or more effectively, leaders should be willing to use it. Preserving work simply because people have historically done it is not a leadership objective.

The more useful question is where technology creates leverage, where human involvement creates value, and where the two should work together.

AI is particularly strong at information processing, pattern recognition, continuous monitoring, knowledge retrieval, automation, anomaly detection, and increasingly sophisticated forms of analysis and execution. Humans remain responsible for objectives, values, context, acceptable risk, ethical boundaries, and accountability.

Between those areas is a shared decision space: scenario evaluation, alternatives, risk analysis, decision support, execution, and learning. That boundary will move as technology improves. Responsibility does not.

Decentralization still matters

Complex environments reward decisions made close to the problem, but decentralization works only when authority is accompanied by information and capability.

Effective decentralization occurs when decision authority, decision-relevant information, and capable human judgment converge at the level closest to the problem.

Giving someone authority without information is not empowerment. Giving them information without authority creates delay. Giving them both without developing judgment simply moves the risk downward.

AI can make decentralized organizations substantially more capable because it can place sophisticated information and analysis closer to the point of action. The leadership task is to design the system so that people have the authority, context, and capability to use that leverage well.

A working framework, not a finished theory

Human-in-Complexity is a scholar-practitioner framework. I do not present it as a universal or finished theory. It is an attempt to put structure around patterns that emerged through operational leadership, then test and refine those ideas through scholarship and the realities of AI-enabled organizations.

That distinction matters. A useful framework should be able to survive contact with reality. Where its propositions fail, they should be revised. Where evidence contradicts an assumption, the assumption should change.

The objective is not to defend the framework. The objective is to develop a better way to think about leadership as human beings work inside increasingly capable technological systems.

Core propositions

• Technology should increase human capability without diminishing human responsibility.
• Experience should be an input into judgment, not a substitute for it.
• AI can expand the decision space, while human judgment determines what should be done within it.
• Technology can participate in a decision, but it cannot absorb the leader’s responsibility for the system in which that decision was made.
• Effective decentralization occurs when decision authority, decision-relevant information, and capable human judgment converge at the level closest to the problem.