🎮 Symbolic AI · Lesson 3
📦 The Applications: Putting the Tools to Work
From Bricks to Buildings
In Lesson 2, we mastered the three core tools: Knowledge Representation (storing rules), Search Algorithms (navigating those rules), and Inference Engines (combining rules to deduce new facts).
Now, let's see how these tools are combined to create real-world applications. We will explore several classic applications of Symbolic AI.
For each application, we will start with a scenario. Your job is to figure out which of the three tools is doing the heavy lifting.
🎲 Application A: Game-Playing AI
🎮 The Scenario: Tic-Tac-Toe
An AI plays Tic-Tac-Toe against a human. It builds a massive "tree" of every possible move it can make, then every possible counter-move by the opponent, and so on.
It assigns a score to the end of each branch (Win = +1, Draw = 0, Loss = -1).
The AI assumes the opponent will always pick the move that gives them the best score (minimizing the AI's score).
Therefore, the AI picks the move that maximizes its own score while assuming the worst-case scenario from the opponent.
Which of the three tools dominates here?
This is called: Adversarial Search (Game-Playing)
Why B is correct: The intelligence comes almost entirely from the Search Algorithm exploring the tree of future states.
This specific type of search is called Adversarial Search, usually implemented using an algorithm like Minimax.
⚖️ How it compares to the GPS from Lesson 2
In a GPS map, the world is static; the roads don't change.
In a game, the world is dynamic because an opponent is actively trying to ruin your plans.
Adversarial search must account for this by assuming the opponent is perfect.
🚁 Application B: The Delivery Drone
🚁 The Scenario:
A delivery drone needs to deliver a package. It must plan a sequence of actions: Pick up Package A -> Fly to Zone B -> Drop off Package A -> Fly to Warehouse C -> Refuel.
It uses Knowledge Representation to know that it cannot drop off a package before picking it up (a constraint), and it tries to find the most efficient sequence that satisfies all these constraints to minimize battery usage.
Which tools are dominating here?
This is called: Automated Planning
Why C is correct: The drone needs Knowledge Representation to understand what a "pickup" or "drop-off" means and what actions are allowed. It also needs Search Algorithms to navigate the vast number of possible sequences to find the most efficient one.
🧩 Why this is different from Game-Playing
There is no "opponent" in automated planning. The drone is just dealing with a static, physical world and trying to find the optimal sequence of steps to reach a goal.
🌐 Application C: The Knowledge Graph
🔎 The Scenario: Google's Knowledge Panel
When you search for "Leonardo DiCaprio," a box pops up on the right showing he is an Actor, was born in 1974, and starred in Titanic.
The AI uses a massive graph of nodes and edges to link these concepts together, allowing it to understand the meaning behind your query, not just the exact text of your search.
Which tool is dominating here?
This is called: The Semantic Web / Knowledge Graphs
Why A is correct: This application is a pure showcase of Knowledge Representation. The intelligence lies entirely in how the data is structured and linked using a Semantic Network. It does not need to infer new facts or search through future states.
🩺 Application D: Expert Systems (The Flagship)
Let's review what we've learned:
- Game-Playing: Pure Search (assuming an opponent).
- Delivery Drone: Search + Knowledge Representation (finding a sequence).
- Knowledge Graph: Pure Knowledge Representation (linking concepts).
Now, let's look at the most well-known, comprehensive, and historically important application of Symbolic AI: the Expert System.
An Expert System packages all three tools into a single software program designed to mimic a human expert in a specific field.
💻 The Example: The IT Support "TechBot"
The system holds a massive database of rules (Knowledge Representation).
When a user reports "My computer is slow," the system uses a Search Algorithm to quickly ignore irrelevant rules (like printer problems) and focus on the "CPU" subset.
Then, the Inference Engine chains rules together:
- Rule 1: IF (CPU usage > 90%) THEN (Cause is a runaway process).
- Rule 2: IF (Cause is a runaway process) AND (Process is Antivirus) THEN (Recommend scheduling the scan).
The system outputs: "Your computer is slow because your antivirus is running a heavy scan. Please schedule it for 2:00 AM."
🧠 Why it is the "Flagship":
Unlike the other applications that dominate with one or two tools, Expert Systems use ALL THREE perfectly: storing rules, navigating them, and logically chaining them together to solve complex problems.
🎯 Final Check: Identify the Application
Now, I will give you a real-world scenario, and you must identify which specific Symbolic AI Application is being used.
Why B is correct: This is a classic Automated Planning problem. It requires finding a sequential set of actions (Clean Kitchen -> Clean Living Room -> Return to Dock) to achieve a goal while satisfying constraints (like not running out of battery). It is not against an opponent (A), doesn't just store relationships (C), and isn't diagnosing a problem based on rules (D).
Why A is correct: This is the purest definition of Adversarial Search. It assumes a hostile, perfect opponent and uses minimax search to find the optimal move.
Why C is correct: This is the textbook definition of a Semantic Network or Knowledge Graph. It is purely Knowledge Representation in action—showing the relationships between different nodes of data.
Why D is correct: This is the quintessential Expert System. It uses all three tools: a massive rule base (Knowledge Representation), an informed search to ignore irrelevant rules, and an inference engine to chain the relevant rules together to reach a diagnosis.
🎉 You've mastered the Applications of Symbolic AI!
Final Recap:
• Adversarial Search: Pure Search (Games).
• Automated Planning: Search + KR (Robots/Drones).
• Semantic Web: Pure KR (Search Engines).
• Expert Systems: All Three Tools combined (Diagnosis).
This wraps up our deep dive into Symbolic AI! In the next module, we will explore the completely different world of Machine Learning—the "Black Box" approach.