Introduction

“We humans are a causal bunch. We want to know why things happen. We want to know why something, or someone caused something else to happen. And we want to know why things turn out the way they do.” 

In the pursuit of truly intelligent process engineering, understanding not just what happens, but why it happens, is essential for transparency, auditability, and trust in AI-driven systems. Human curiosity about causality—why things happen—has shaped philosophy, science, and now artificial intelligence. Yet, as Judea Pearl’s “Ladder of Causation” and philosophical traditions remind us, correlation does not imply causation. To move beyond the “post hoc ergo propter hoc” fallacy, we must equip AI with the ability to reason over causes, interventions, and counterfactuals.

This is particularly true when implementing Smart Monitoring frameworks where the data stream needs to be correctly ascertained to obviate noise so that predictions can be correct….

The Human Fascination with Causality

Humans are innately driven to seek causal explanations for events. From Aristotle’s four causes (material, efficient, formal, final) to the DIKIW (Data–Information–Knowledge–Intelligence–Wisdom) hierarchy, our understanding of the world is built on layers of abstraction and meaning.

A long tradition in psychology and philosophy is investigating the principles of causal understanding. Causality governs the relationship between events. Formalizing this, the world consists of a collection of causal systems; in each causal system there is a set of observable causal variables. Causal systems are observed on a set of trials—on each trial, each causal variable has a value. Most ancient and middle century philosophers took a view that cause is behind every phenomenon. In ancient Greece, Aristotle believed that cause involved four features:

1. Material cause: The most important or effective agent in producing a certain effect. For example, the clay is the material cause of the pot.
2. Efficient cause: The immediate agent of a change or the use of a cause to bring about a desired result. For example, the potter is the efficient cause of the pot.
3. Formal cause: The essential properties of a substance, which determine the substance’s behavior under a given set of conditions. For example, the potter shape is the formal cause of the pot.
4. Final cause: A goal or purpose of a thing, the end to which a process is directed. For example, the use of a piece of clay is the final cause of a pot.

In the Aristotelian view, the four causes are different aspects of one underlying principle. In this view, a causal explanation’s goal is to identify the fundamental cause of a thing’s existence.

Phenomenology or Narrative Inquiry is a philosophy and a methodology not limited to an approach to knowing. It is rather an intellectual engagement in interpretations and “meaning making” that is used to understand the lived world of human beings at a conscious level. Phenomenological approaches are more effective in describing, rather than explaining subjective realities in terms of insights, beliefs, motivation, actions and wisdom.

Phenomenology and narrative inquiry highlight that meaning-making is as much about subjective experience as it is about objective explanation. However, our reasoning is often vulnerable to the “after this, therefore because of this” fallacy, where we mistake sequence for consequence.

Pearl’s Ladder of Causation

Judea Pearl’s “The Book of Why” fundamentally reshapes our understanding of causality by introducing a formal framework—particularly the Ladder of Causation—that enables AI systems to move beyond correlation towards true causal reasoning. This paradigm shift empowers AI to perform interventions, reason about counterfactuals, and generate explanations, which are critical for building transparent, trustworthy, and adaptive enterprise systems. Companies like CausaLens are actively developing tools and platforms that embed Pearl’s causal inference models, such as structural causal models (SCMs) and causal graphs, to enhance decision-making, predictive analytics, and automation in industries ranging from healthcare to manufacturing, aligning with the broader goal of creating intelligent systems capable of understanding and acting upon cause-effect relationships in complex environments.

Judea Pearl’s framework divides causal reasoning into three ascending rungs:

1. Association (Seeing):

   • What is?

   • Observing statistical relationships and correlations.

   • Example: “What does a sensor reading tell me about machine health?”

2. Intervention (Doing):

   • What if?

   • Predicting the effects of deliberate actions or changes.

   • Example: “What if I adjust the pressure—will yield improve?”

3. Counterfactuals (Imagining/Retrospection):

   • Why?

   • Reasoning about alternative realities and root causes.

   • Example: “Would the defect have occurred if we had used a different material?”

While most AI operates at the first rung, intelligent process engineering demands climbing higher—enabling machines to plan, explain, and learn from interventions and counterfactuals.

Key Concepts from The Book of Why:

1. The Ladder of Causation

   • Rung 1: Association

     • “What is?”

     • Observing correlations (e.g., “Ice cream sales correlate with drownings”).

     • Dominates current AI (e.g., deep learning models).

   • Rung 2: Intervention

     • “What if?”

     • Predicting outcomes of actions (e.g., “What if we ban ice cream sales?”).

     • Requires causal models to simulate interventions.

   • Rung 3: Counterfactuals

     • “Why?”

     • Imagining alternative realities (e.g., “Would the patient have survived with a different treatment?”).

     • Essential for explainability and ethical AI.

2. Limitations of Data-Centric AI
   Pearl critiques the “data-centric” mindset, arguing that causal questions cannot be answered by data alone. For example, observational data might show that students who attend tutoring score higher on tests, but only causal reasoning can determine if tutoring causes better performance or if motivated students self-select into tutoring.

3. Causal Models
   Pearl advocates for structural causal models (SCMs) and directed acyclic graphs (DAGs) to encode cause-effect relationships. These tools enable AI to reason about interventions and counterfactuals, moving beyond pattern recognition to true understanding.

How Companies Like “CausaLens” Are Applying These Ideas

Companies such as CausaLens, Causely, and others are pioneering causal AI solutions that align with Pearl’s vision:

1. CausaLens

   • decisionOS Platform: Enables enterprises to build causal models for root-cause analysis, scenario planning, and decision optimization.

   • Use Cases:

     • Marketing: Identifying the true impact of ad spend on sales (separating causation from correlation).

     • Manufacturing: Pinpointing root causes of defects in production lines.

   • Methods: Combines causal discovery (inferring causal graphs from data) with structural causal modeling.

2. Causely

   • Observability Platform: Uses causal AI to automate IT incident management by mapping application dependencies and inferring root causes of failures.

   • Example: Integrates with Grafana dashboards to enrich alerts with causal analysis, reducing false positives and accelerating troubleshooting.

3. Xplain Data

   • Focuses on causal inference for business decisions, helping companies optimize pricing, inventory, and customer retention by modeling cause-effect relationships.

Implications for AI Development

• Transparency & Trust: Causal AI systems can explain why they make recommendations, addressing the “black box” problem.

• Robust Decision-Making: By understanding interventions (Rung 2) and counterfactuals (Rung 3), AI can simulate outcomes and avoid harmful actions.

• Ethical AI: Causal models help identify and mitigate biases (e.g., distinguishing between correlation and discrimination in hiring algorithms).

Challenges

• Data Requirements: Causal inference often requires experimental data (e.g., A/B tests) or strong domain expertise to build accurate models.

• Complexity: Industrial systems involve high-dimensional, dynamic interactions, making causal graphs difficult to construct and validate.

Pearl’s work laid the theoretical foundation for causal AI, and companies like CausaLens and Causely are turning these ideas into practical tools. By climbing the Ladder of Causation, AI systems can move beyond brittle correlation-based predictions to robust, explainable, and ethical decision-making—a critical step toward building existential intelligence (systems that understand why they act). As Pearl writes, “You are smarter than your data”—and with causal AI, machines can be too.

Causal Inference in Industrial Workflows

Why Causality Matters

• Transparency:
  Causal models can explain why a process failed or succeeded, not just that it did—building trust and enabling root-cause analysis.

• Auditability:
  Regulatory and safety-critical environments require systems to justify decisions, especially when interventions have significant consequences.

• Optimization:
  By understanding true cause-effect relationships, AI can recommend interventions that reliably improve outcomes, rather than chasing spurious correlations.

Practical Applications

• Predictive Maintenance:
  Moving from “machines with high vibration often fail” (association) to “increasing lubrication reduces failure risk” (intervention), and finally to “would this machine have failed if maintenance had been performed?” (counterfactual).

• Process Control:
  Using causal graphs to model the impact of process variables, enabling scenario planning and robust automation.

• Quality Assurance:
  Explaining defects not just with data, but with causal logic—empowering continuous improvement and compliance.

Merging Narrative, Phenomenology, and Causal AI

The article “After this, therefore because of this…” emphasizes that true intelligence—human or artificial—must grapple with both the complexity of cause and effect and the richness of lived experience. Phenomenology and narrative inquiry remind us that not all meaning is captured in data or logic; context, story, and subjective understanding are crucial. For AI to be robust and actionable, it must integrate narrative intelligence with causal models, becoming less brittle and more rational, able to handle both the known and the unknown.

Building Causal AI: From Data to Interventions

To operationalize causal reasoning, intelligent process engineering must:

• Model Causal Structures:
  Use directed acyclic graphs (DAGs) or structural causal models (SCMs) to encode relationships between variables.

• Design Experiments:
  Implement A/B testing, randomized interventions, or natural experiments to validate causal hypotheses.

• Leverage Counterfactual Analysis:
  Simulate alternative scenarios to answer “what if” and “why not” questions—critical for post-mortems and future planning.

Challenges and Opportunities

• Data Limitations:
  Observational data alone is insufficient for higher-rung reasoning; interventions and domain expertise are often required.

• Complexity:
  Industrial systems are high-dimensional and dynamic, making causal modeling non-trivial.

• Explainability:
  Causal models offer a path to more interpretable, auditable AI—aligning with regulatory and ethical imperatives.

Conclusion: Toward Existential Intelligence

As AI matures from pattern recognition to causal reasoning, intelligent process engineering stands to benefit from systems that don’t just predict, but understand and explain. By embracing Pearl’s Ladder of Causation and integrating narrative, phenomenological, and cybernetic perspectives, we can build AI that mirrors human-like cause-effect reasoning—enabling transparent, auditable, and truly intelligent industrial workflows. This is a step toward “existential intelligence”: systems that are not only smart, but wise, ethical, and context-aware.

References:

• Pearl, J. (2018). The Book of Why: The New Science of Cause and Effect.

• Uchil, N., & Vinekar, S. S. (2021). After this, therefore because of this…

• NUMO-EC Series & MSON Series Context

NITIN UCHIL Founder, CEO & Technical Evangelist
nitin.uchil@numorpho.com