“The world around us seems unpredictable because it is complicated” – Anonymous

The world around us may seem chaotic, but we can manage this chaos strategically with foresight, leveraging chaos theory powered by AI. Both in the world of geopolitics as well as business. This was the introductory theme in the first part of this article. Randomness, as per chaos theory, is more “imagined” than real, arising from the ignorance of the causes behind the event.

The first part introduced that we can leverage AI as an advisor to understand the various scenarios of chaos that may originate from an event. We can also use AI to understand why chaos happened. In this second part, we explore some foundational concepts and then get into how AI can help transform a complex analysis like this into self-service tools.

System dynamics

Though system dynamics is not the core underlying principle behind chaos theory, it is a foundational concept for explaining how, sometimes, simple systems can show “weird” behavior. An evolution of system dynamics, business dynamics can be used to explain many baffling business outcomes and help pinpoint the drivers that led to the outcome. What I love about systems or business dynamics is the simplicity of the method. It is simple, yet it can help explain the complex. You can map complex behaviors through fundamental levers like feedback loops, stock, flows, etc..

Dynamical system theories

Dynamical system theory uses advanced methods (read mathematics) but with the same motive-understand the behavior of complex systems. Dynamical system theory is leveraged to understand the behavior of complex systems (dynamical systems) by employing differential equations. Chaos theory is a dynamical system theory.

Chaos theory

Even with most deterministic systems, randomness seems inevitable. In other words, chaos can be seen even in bodies or events governed by deterministic laws. Chaos theory studies deterministic chaos. And as the term “deterministic chaos” indicates, it helps link two terms considered exact opposites. The primary postulate is that what seems like chaos is essentially our ignorance of root causes or behaviors.

Though the key underlying principles of chaos theory are mathematical, like the system must be topological transitive and should have dense periodic orbits, the traits can be summarized in plain English. So, in much simpler terms, chaos theory postulates that embedded within the randomness of complex, chaotic systems, we can find

• Repetition,
• Patterns,
• Self-organisation
• Interconnectedness,
• Self-similarity
• Constant feedback loops. 

This allows us to mathematically model what is considered complex. It does not matter whether it is business or geopolitics. You can model systems that seem chaotic and, hence, predict their behavior. Predictability of what seems unpredictable is, hence, the core capability of these systems.

But this is not new and has been done for decades. What is new, however, is how AI can help put something like this, traditionally used by mathematicians and technologically proficient, in the hands of leaders and executives who can use this as a “self-service” tool.

Diagnostic and Predictive

As you can tell at this point, chaos theory finds its applications in diagnostic and predictive analytics. For example, it can help explain random behaviors in physics (diagnostic) but can also help predict weather. Examples of areas that leverage chaos theory for diagnostic and predictive analytics are computer science, geology, engineering, meteorology, physics, population dynamics, robotics, biology, anthropology, mathematics, politics, philosophy, and economics. However, the analysis is mainly done by scientists. In areas like politics and business, it will be great if we can put this tool’s power into the practitioners’ hands. And this is where AI can help.

Self-service strategy

Let us start with a simple example. Consider a straightforward chaos theory analytics approach- recurrence plots. You plot data in a certain way, and the visual appearance of the data, the plot, can tell you a lot about the system’s dynamics. We do this frequently in industry, with less chaotic time series data. Recurrence plots do this at a more advanced level with chaotic systems, but you can see where I am going.

Some tools can analyze time series and identify patterns. You can leverage the same approach to develop tools that can leverage AI algorithms to identify patterns and anomalies, pinpointing behaviors that lead to chaos. However, the recurrence plot is one of the simplest examples. However, the good news is that both tech infrastructure and algorithms exist to “automate” chaos theory analytics for complex methods.

The example I used in the first part of the article, the current crisis in Middle East, is a simple application of chaos theory. It falls more within the realm of dynamical systems than pure chaos theory. A version of the algorithm I suggested in this article, can help predict the system’s behavior. The approach will be similar to how we forecast weather, with two key differences. One, the predictability will be much better. Second, it will be intuitive and interactive for the leader or executive. Simulation has been used in war gaming for decades. AI can make it much more granular and accurate and extend it to include political behavior impacts.

An advantage, as highlighted in the very heading itself, of using these tools is that the recommendations are devoid of emotions. If you ask a military general, hundreds of whose people have been massacred by a terrorist organization early that morning, what the next steps should be, what do you think will be the response? We, humans, are governed significantly by our emotions. AI is not. While you can design AI algorithms to “mimic” emotions, creativity etc. , remember that it is all design. The AI-driven apocalypse will not happen because AI “decided” something. It will happen because a human-designed a flawed algorithm.

The advantages in the business world, where gigantic heaps of data exist for these models, are enormous. If we go back to the pandemic, a model designed to detect the “butterfly effect,” like the initial lockdowns in China, could have predicted repercussions. This is not the best example, though, since COVID was a black swan event, so comparable past behaviors may not have been available. But a model could have still helped anticipate disruptions. But let us get into the more tactical applications.

Let us go back to the recurrence plot. A vast amount of time series data exists in supply chains. This data hides the deterministic behavior of complex supply chains, many of which we perceive as random or chaotic. Humanly, we can’t take a big-picture view of all that data by leveraging dynamic system models. However, deep learning can help make sense of that in many different ways. We will cover some in a separate article.

To conclude, the key benefits of leveraging AI in the context of applied chaos theory in business and politics will be:

• Look at the bigger picture by analyzing a vast amount of data
• Help both diagnose as well as predict
• Put the power of dynamic system modeling into the hands of citizen analysts, executives and leaders

References:

• Chaos and Dynamical Systems- Princeton University Press
• Chaos Engineering (Manning Publication)- Mikolaj Powlikowski
• Chaotic systems in digital communications (CRC press) -Marcio et al
• Creative management of complex systems (Wiley) -Heraud et al
• Market Entropy (Business Expert Press)- Rajagopal