Why Unmodeled Decisions Are Becoming the Biggest Risk in Business

Abstract

For decades, organizations have described themselves as data-driven. Yet the most consequential business decisions have rarely been driven by structured data alone. They depend on conversations, judgment, competitive dynamics, customer sentiment, organizational behavior, and countless other forms of qualitative knowledge that resist traditional quantitative analysis.

The emergence of foundation models changes this equation.

Their greatest economic contribution is not generating text, writing code, or answering questions. It is dramatically reducing the cost of transforming unstructured human knowledge into structured decision variables. That seemingly technical capability changes the economics of decision modeling itself.

As decision modeling becomes dramatically less expensive, relying on unmodeled decisions becomes increasingly difficult to justify. In the coming decade, the greatest business risk may no longer be uncertainty itself — but choosing not to model it.

For more than two decades, business leaders have repeated the same aspiration: we want to be data-driven.

#The illusion of being data-driven

The phrase became so common that it gradually evolved into management doctrine. Organizations invested billions of dollars in data warehouses, business intelligence platforms, predictive analytics, and machine learning. Dashboards appeared in every executive meeting, and every major decision was expected to be supported by data.

And yet something interesting happened.

Despite having more data than at any point in history, many of the most important business decisions still looked remarkably similar to those made twenty years ago. Questions such as Should we enter this market? Is this competitor becoming a real threat? Can we trust this partner? Has this technology reached an inflection point? rarely have clean numerical answers. They emerge from conversations, observations, experience, intuition, and countless weak signals scattered across an organization.

Those decisions are qualitative by nature, yet they determine billion-dollar investments.

This led me to question what “data-driven” had really meant.

Perhaps organizations were never truly driven by data. They were driven by structured data.

#Business has never lacked information — only measurable information

Organizations have never suffered from a shortage of information.

They are surrounded by customer conversations, product feedback, recruiting trends, competitive intelligence, partner discussions, regulatory developments, and years of accumulated executive experience. The challenge has never been collecting this information. The challenge has been transforming it into something that economic and strategic models can consume.

That distinction may sound subtle, but I believe it represents one of the defining economic consequences of foundation models.

Business has never lacked information. It has lacked measurable information.

#Every decision already has a model

Every executive already operates with a model of the business.

When someone says, “I have a gut feeling,” they are not making a decision without a model. They are relying on a model that exists only in their own mind. Experience has encoded years of observations into an internal representation of how markets behave.

The problem is not intuition.

The problem is that hidden models cannot be inspected, challenged, stress-tested, transferred, or systematically improved. They remain personal rather than organizational.

The future is not about replacing intuition.

It is about making intuition modelable.

#Experience-driven, data-driven, model-driven

Business decision making has quietly evolved through three eras.

The first was experience-driven. Competitive advantage depended largely on individual judgment. The model lived in the executive’s head.

The second became data-driven. Organizations learned to measure what fit naturally into databases and dashboards. Analytics dramatically improved decisions, but only for the information that was already structured.

The third era is beginning now: model-driven decision making.

Instead of asking What data do we have?, organizations can increasingly ask What assumptions should we model?

That is a much deeper shift.

#The first-order effect of foundation models

Much of today’s discussion about AI focuses on productivity. Can it write software? Summarize documents? Generate presentations? These are valuable capabilities, but I suspect they are not the most important ones.

The deeper contribution is that foundation models dramatically reduce the cost of translating unstructured human knowledge into structured decision variables.

For decades, management science has already provided extraordinary frameworks: Porter’s Five Forces, game theory, Bayesian decision theory, systems thinking, scenario planning, and Monte Carlo simulation. None of these were waiting for new mathematics.

They were waiting for better inputs.

The expensive part was never running the model. The expensive part was transforming messy business reality into something the model could consume.

That translation process traditionally required teams of analysts, consultants, interviews, workshops, and months of synthesis. Much of it remained too expensive to perform consistently, which is why many strategic decisions continued to rely on executive intuition.

Foundation models collapse much of that translation cost.

That changes the economics of management itself.

#The missing layer

Most conversations about enterprise AI focus on the top and bottom of the technology stack — foundation models, and the applications built on them.

That view misses the layers where much of the economic value is actually created.

Foundation ModelsInterpret language and context
Information StructuringConvert fuzzy signals into structured variables
Decision ModelingApply economics, game theory, simulation, and optimization
Human JudgmentChallenge assumptions and choose among trade-offs
Business ValueImprove resource allocation under uncertainty

The highlighted middle layers — Information Structuring and Decision Modeling — represent the missing bridge between foundation models and business value.

Foundation models enable information structuring. Information structuring converts customer conversations, market observations, competitive signals, and institutional knowledge into structured variables. Those variables can then enter decision models built from economics, game theory, scenario analysis, simulation, and optimization.

The output of these models is not an automatic answer. It is a more explicit representation of assumptions, trade-offs, probabilities, and possible outcomes. Human judgment remains essential, but it now operates on top of a model that can be inspected, challenged, and improved.

The greatest economic value therefore does not emerge from generating language alone. It emerges from transforming language into models that improve decisions.

Language becomes mathematics.

#Decision Capital

This realization leads to a broader idea that I call Decision Capital.

Just as financial capital enables investment and human capital enables execution, Decision Capital represents a distinct organizational asset. It is built through better models, stronger feedback loops, institutional learning, and the ability to integrate qualitative knowledge into rigorous reasoning.

Foundation models do not create Decision Capital.

They dramatically accelerate its accumulation.

Organizations that consistently build better decision models will gradually outperform those that rely primarily on fragmented intuition.

#Why unmodeled decisions become a strategic liability

Business has always operated under uncertainty. That has never changed.

What is changing is our ability to reduce unnecessary uncertainty.

For decades, many strategic decisions remained unmodeled because constructing rigorous models required too much time, expertise, and manual effort. That was a rational compromise.

It is becoming a less rational one.

As the cost of decision modeling continues to fall, failing to model important decisions is no longer simply a limitation. It becomes an avoidable source of risk.

The greatest business risk is no longer uncertainty itself. It is unmodeled uncertainty.

#Looking forward

This essay introduces the first building block of a broader research agenda called Computational Economy.

If the Industrial Revolution amplified physical labor, and the Internet amplified access to information, then foundation models may ultimately be remembered for something more fundamental.

They change the economics of human judgment.

The essays that follow will develop this idea further through the concepts of Decision Capital, Computational Organizations, and The Economics of Foundation Models, with the goal of building a broader framework for understanding how computation is reshaping business, science, and society.