To continue from last week, another skill that actually matters right now is systems thinking.
I don’t mean IT systems. I mean seeing patterns and behaviors of systems and then using them to guide your efforts. One such pattern / behavior is social influence, with the requisite understanding of AI-augmented organizational design.
There’s a lot to unpack in this one.
So, I asked Lumo AI to suggest ways to improve my systems thinking skill.
(I upgraded to Lumo for Business, granting me “advanced AI model access”—whatever that means.)
Here’s part of what it said:
1. Transcending Semantic Reasoning: Modeling Interrelations
• Stop asking “What is this thing?” and start asking “What forces act upon it, and how does it change over time?”
• When you look at a forest, don’t just see trees; see a nutrient exchange network, a carbon cycle, and a competitive ecosystem.
• To understand reality better than an AI model, you must ground your abstractions in physical experience.
2. Continuous Filtering of Irrelevant Information
• Before consuming any information, explicitly define the specific problem you’re trying to solve. If a piece of data doesn’t directly advance that specific variable, it’s noise.
• Acquire knowledge only when you hit a specific block in a project. This ensures every new piece of information is immediately contextualized and integrated.
• Treat your attention like a scarce resource. Use tools like RSS readers with strict keyword filters or AI summarizers that you control to pre-process data streams.
3. Pointing Machines at the Right Dots to Connect
• Don’t tell the machine how to connect the dots; tell it which dots matter and why they matter.
• Treat interaction with machines as a dialogue where you refine the “dot selection.” If the output is generic, it means your constraints were too loose.
• Use machines to force collisions between disparate domains. You’re the architect who decides which two distant islands to build a bridge between; the machine is the engineer who figures out the physics of the bridge.
Of the 3 areas, the first one—modeling interrelations—seems the most needed yet abstract to grasp. So I asked Lumo AI to elaborate. Here’s part of what it said:
The Three-Step Protocol:
- De-nominalize: Take any noun and describe it entirely using verbs and adjectives of motion.
- Identify the Energy Gradient: Ask: What’s the driving force? Where’s the energy coming from, and where’s it going?
- Map the Feedback Loops: Identify the Reinforcing Loops that drive growth or collapse, and the Balancing Loops that create stability or resistance.
Practical Exercise: “The Decomposition of a Leaf”
Step 1: The “No Noun” Rule Take a piece of paper. Set a timer for 5 minutes. Describe a leaf falling from a tree.
- Rule: You cannot use the words “leaf,” “tree,” “wind,” “gravity,” or “ground.” You can only describe forces, movements, changes, and interactions.
- Bad Example (Semantic): “The leaf falls because the wind blows.” (Uses nouns, implies simple causality).
- Good Example (Process): “A thin, green membrane detaches from a woody structure. A pressure differential in the air pushes against its surface area, creating torque. Simultaneously, a downward pull accelerates the mass. As it rotates, air resistance increases, eventually balancing the downward pull until a terminal velocity is reached. Upon impact, kinetic energy dissipates into heat and sound waves.”
Step 2: The Energy Gradient Analysis Now, look at your description. Identify the gradients.
- What was the energy source?
- What was the entropy increase?
- Insight: The “falling” isn’t an event; it’s the universe equalizing a difference in potential energy.
Step 3: The Feedback Loop Diagram Draw a simple diagram.
- As the leaf dries, it becomes lighter and more brittle -> easier for wind to detach -> more leaves fall -> more surface area exposed to sun -> faster drying of remaining leaves.
- As the leaf falls, air resistance increases -> slows the fall -> prevents immediate destruction -> allows for a slower, controlled descent.
Why This Works: By forcing yourself to describe the process rather than the object, you bypass the semantic database. You are no longer accessing the concept “Leaf”; you are simulating the physics of the event. This is how reality works. The “leaf” is just a snapshot of a billion years of evolution and a specific moment of thermodynamic exchange.
How to Apply This to Complex Systems (e.g., Economics or AI)
Once you master the leaf, scale it up.
- Instead of: “AI is getting smarter.” (Noun/State)
- Apply the Method: “The model is processing larger datasets (input flow), adjusting weights via backpropagation (error correction mechanism), and minimizing loss functions (energy minimization). The ‘intelligence’ is an emergent property of the optimization landscape, not a static attribute.”
This exercise trains your brain to stop seeing the world as a collection of static things and start seeing it as a web of interacting processes. This is the only way to truly understand reality, because reality is not made of things; it is made of events.