Mengenal Artificial Intelligence (Kecerdasan Buatan) - Kuliah AI #01

Main Ideas / Concepts Covered

• Artificial Intelligence (AI) is not limited to robots or self-driving cars.
  • AI is described as everywhere around us in daily life.
  • Examples are used to show AI in common technologies and games.

Everyday Applications of AI (Illustrated Examples)

• Google Maps: uses AI techniques to provide the fastest route (search/recommendation).
• Video games (e.g., Mobile Legends, PUBG, Dota, Counter-Strike): game AIs use algorithms so bots/enemies can:
  • find routes
  • decide tactics (e.g., when to shoot, when to hide)
  • behave “smartly” to challenge the player
• Email filtering (e.g., spam detection):
  • AI can distinguish spam/fraudulent/deceptive emails from legitimate ones.
• Smartphone biometrics:
  • Face Unlock and fingerprint scanners recognize the owner by comparing facial or fingerprint patterns.
• Voice assistants (e.g., Google Voice Assistant, Cortana):
  • can recognize speech.
• Photo recognition (Facebook):
  • can recognize faces in uploaded photos.
• Social media filters (Instagram/TikTok):
  • computer vision tracks facial landmark positions (e.g., nose/ears aligned to the user’s face).
• Driver safety / monitoring:
  • a computer camera can read license plates (linked to surveillance/tracking).
  • includes a cautionary mention about not wearing seat belts/helmets.
• Smart home appliances:
  • AI is implied in devices like washing machines, rice cookers, robot vacuums, AC, etc.

Definition of AI Evolves Over Time

The lecture presents an evolving definition across decades:

• 1970s (initial idea): AI as a system that can think like humans.
• 1990s (updated): AI must be able to act/do work like humans (not just think).
• Early 1990s (refinement): AI defined as a system that thinks rationally.
  • Reasoning: defining AI as “human-like” can be less appropriate because humans can act irrationally due to emotions and feelings (AI doesn’t have these).
• Later refinement (still relevant by the lecturer’s version):
  • AI is defined as a system that can act rationally and work rationally.

Core Techniques of AI (4-Part Framework)

The lecture summarizes AI into 4 basic techniques:

1. Searching
2. Reasoning
3. Planning
4. Learning

It notes that each technique includes subtypes/method families, such as:

• Searching: systematic search, probabilistic search, etc.
• Reasoning: rule-based, fuzzy-logic-based, certainty-based, etc.
• Learning: machine learning with supervised, unsupervised, and reinforcement learning.

---

Methodology / Instructions Presented

A) AI Technique 1: Searching

• Purpose: Find solutions by systematically searching for steps to reach a goal.
• What it does: produces an ordered path/sequence (e.g., routes).
• Example: Google Maps route planning (from starting location to destination using systematic steps).

B) AI Technique 2: Reasoning

• Purpose: Draw conclusions from data/facts.
• What it does: uses known information to infer likely outcomes.
• Example: an expert system for diagnosing corn diseases:
  • user inputs observed symptoms
  • system concludes which disease it is
  • system provides recommendations for treatment

C) AI Technique 3: Planning

• Purpose: break down a complex problem into smaller simpler subproblems.
• What it does: determines what to do first, then next steps, etc.
• Example: elevator control/planning:
  • plan which floors to go to based on button presses
  • determine the order of movement (e.g., who goes up first, who goes down first, and subsequent moves)

D) AI Technique 4: Learning

• Purpose: teach the computer to recognize patterns in data so it can make decisions.
• What it does: learns facial patterns for Face Unlock and fingerprint patterns for unlocking.
• Subtypes:
  • Supervised learning
  • Unsupervised learning
  • Reinforcement learning
• Examples:
  • Face Unlock:
    • first scan teaches the facial pattern (e.g., nose position, eyes, eyebrows, etc.)
    • later it matches the live scan to unlock
  • Fingerprint unlocking:
    • multiple scans teach/confirm fingerprint shape/pattern for recognition

---

Key Lessons / Takeaways

• AI is a broad field used in many everyday systems—not just robots.
• AI definitions have shifted toward rational decision-making and action.
• A practical mental model of AI is:

Searching + Reasoning + Planning + Learning

• Among them, learning/machine learning is emphasized as especially widely discussed today.

---

Speakers or Sources Featured

• Speaker: Ahmad Fahri (also mentions “Muhammad Fahri” when signing off).
• Books / authors referenced (sources):
  • A large AI book described as 1,000+ pages, credited to “Suyanto” (referred to as a lecturer’s book).
  • Another AI book locally available described as about 200 pages, also attributed to Mr. Suyanto in the subtitles.
• Named technologies / platforms (mentioned examples, not sources):
  • Google Maps, Google Voice Assistant, Cortana
  • Facebook, Instagram, TikTok
  • Mobile Legends, PUBG, Dota, Counter-Strike
  • Email/spam filters
  • Smartphone Face Unlock / fingerprint scanners
  • Social/face filters (e.g., rabbit nose/ears)