The Path of Least Reflection

The Difference Between Doing the Work and Understanding It

Enterprise software is evolving from systems that help people perform work toward systems increasingly capable of performing that work themselves.

This is an exciting development. Many activities that consume valuable human time can and should become simpler.

There is a distinction, however, between completing an activity and developing the understanding necessary to determine whether that activity is appropriate in the first place.

Consider workforce planning.

An AI-enabled system might analyze workforce demand, identify staffing gaps, compare scenarios, and recommend adjustments. It could potentially initiate approved actions based on those recommendations.

The analysis might be accurate. The transactions might be executed correctly.

Yet an experienced operational leader may know that a particular team's productivity depends on informal mentoring relationships that aren't reflected in the available data. An HR leader may recognize that a seemingly efficient staffing recommendation could undermine a capability the organization has spent years developing.

Those perspectives are important.

They may change the interpretation of the analysis, reveal an alternative, or lead to an entirely different question.

The value of human participation isn't limited to checking whether the system reached the right answer. Sometimes human participation changes what the organization understands the question to be.

That is a fundamentally different contribution.

Interaction Is Not Necessarily Collaboration

As enterprise AI evolves, we're seeing more opportunities for people to interact with intelligent systems.

Users can ask follow-up questions, request additional analysis, refine recommendations, and explore alternatives.

These capabilities are valuable. They may also provide a foundation for something more significant.

Interaction alone does not establish collaboration.

A person can ask an AI system ten questions without contributing meaningful knowledge or developing a deeper understanding. Similarly, a person can approve an AI-generated recommendation without genuinely exercising judgment.

The existence of a human approval step doesn't necessarily mean the human has participated in the thinking.

Interaction is a capability. Collaboration is a design choice. Knowledge emergence is a potential outcome.

Meaningful collaboration occurs when human and artificial intelligence contribute complementary capabilities to a shared process of inquiry.

People bring experience, context, judgment, values, and knowledge that may never have been formally documented.

AI can help analyze patterns, surface relationships, explore alternatives, and challenge assumptions.

Together, they may develop an understanding that neither possessed at the beginning.

The opportunity isn't simply to make AI's answers better.

It's to make the organization's thinking better.

When Knowledge Emerges Through Thinking Together

Much of what people know about their organizations is difficult to find in a database, process document, or system configuration.

It's embedded in experience.

A manager knows why a particular exception exists. An employee understands which informal practices make a process successful. A team remembers why an earlier initiative failed, even though the official project record describes a different problem.

Often, that knowledge becomes visible only when someone has an opportunity to articulate it.

A thoughtful question can reveal an assumption. A comparison can expose a contradiction. An unexpected recommendation can prompt someone to explain something they hadn't previously recognized as important.

The act of articulating knowledge can itself produce new understanding.

This is where I see particular promise in collaborative intelligence.

Imagine an AI-assisted planning process in which employees contribute their experiences, the system helps reveal relationships and alternatives, and participants develop a more complete understanding of the work together.

The outcome may be a better decision. Something else has happened, too.

The organization has learned something it didn't know before.

That learning has value beyond the immediate transaction.

Which raises another question: What happens to the knowledge after the collaboration ends?

If it remains buried in a conversation history, the organization may have experienced a valuable moment of discovery without developing a lasting organizational capability.

Preserving that knowledge requires more than storing a transcript.

It requires a mechanism for identifying what emerged, understanding its context, validating its relevance, and making it available for future decisions.

Capturing information preserves a record. Stewarding emergent knowledge develops an organizational asset.

Knowledge evolves; stewardship must include the ability to question, refine, and sometimes retire what was previously understood.

The Quiet Erosion of Organizational Ownership

Organizations increasingly rely on enterprise software they don't own, delivered through platforms they don't control.

That arrangement isn't inherently problematic. It provides access to sophisticated capabilities that would be impractical for many organizations to develop independently.

The more consequential question arises when those platforms begin performing not only transactions, but also analysis, planning, recommendations, and decisions.

Organizations may retain formal authority over their policies and outcomes.

But ownership involves more than retaining the right to approve a decision or being accountable when something goes wrong.

It also involves the ability to explain why decisions were made, understand the assumptions behind them, challenge alternatives, and adapt when circumstances change.

An organization can retain strategic control while gradually losing the practiced capability to exercise it.

That possibility deserves attention.

The danger isn't necessarily that AI will take control away from organizations.

It's that convenience may gradually reduce the occasions when people exercise the judgment, curiosity, and shared understanding through which organizational ownership is sustained.

Over time, the organization may become exceptionally capable of having work performed for it while becoming less capable of understanding how to improve that work independently.

This isn't an inevitable outcome of agentic AI.

It is a design risk that should be addressed.

Collaboration Must Be Designed Into the Work

Human agency is often discussed as something organizations should preserve.

I agree. Preserving agency requires more than encouraging employees to remain engaged or instructing them to question AI-generated recommendations.

We know that people tend to follow the path of least resistance, particularly when systems are convenient, reliable, and designed to reduce effort.

We cannot rely on human intention alone to preserve meaningful human agency.

We have to design opportunities for collaboration into the work itself.

That doesn't mean every transaction requires a conversation or every automated process needs human approval.

Autonomous execution is entirely appropriate for many activities.

The more thoughtful question is where human participation materially contributes to value.

Where does experience change the interpretation of data?

Where do competing perspectives help reveal assumptions?

Where could an AI-generated recommendation become the starting point for discovery rather than the end of a decision process?

Where should people remain actively involved because the organization needs to preserve and develop its own judgment?

Lastly, how will the knowledge gained through those exchanges become available to others?

These questions should influence how organizations design workflows, decision rights, delegated authority, learning mechanisms, and measures of success.

Collaboration shouldn't depend on an employee remembering to ask one more question after the system has already delivered a convincing answer.

It should be intentionally supported at the points where thinking together creates meaningful value.

From Capability to Intentional Design

This is why I believe organizations need to begin with value rather than technology.

First, understand what the organization is trying to achieve and what capabilities it needs to develop or preserve.

Then determine how human intelligence, artificial intelligence, and collaboration can best contribute to that value.

Only then design the work, including where autonomy is appropriate, where oversight is necessary, and where collaboration creates the greatest opportunity.

Value → Intelligence → Design.

The sequence is critical to ownership.

If organizations begin with what technology can automate, they may inadvertently allow available capabilities to shape the work before fully considering what the work should become.

If they begin with value, they have a better opportunity to design technology-enabled work that strengthens both performance and organizational capability.

The question isn't whether AI should do more. In many circumstances, it should.

The more consequential question is whether organizations are being equally intentional about what people should continue to understand, contribute, discover, and own.

A Question Worth Pondering

We are entering a period in which increasingly sophisticated technology may make execution remarkably easy.

That creates a tremendous opportunity to redirect human effort toward deeper thinking, better decisions, and new forms of collaboration.

That outcome won't happen simply because people have more time.

It will depend on what we design that time, intelligence, and participation to accomplish.

When technology makes execution effortless, reflection must become intentional, and collaboration must become part of the architecture of work.

Perhaps one of the most important measures of successful AI-enabled transformation will be this:

Has the organization become more capable of understanding, directing, and improving its own work, or merely more capable of having that work performed for it?

The organization should remain the author of its transformation.

HR SOS Leadership Series | Sustainably Optimized Success