🧠 History of AI

1. Classic Programming vs. Machine Learning

   In traditional programming, you provide data + program → output. In machine learning, you flip the flow: you provide data + output → program. The learned “program” (model) can then take new data and generate predictions — a process known as testing.

2. Minimax Algorithm & Checkers

   The minimax algorithm was one of the earliest AI algorithms used in games like checkers. It famously beat the world’s best checkers player before he passed away by cancer — a milestone in demonstrating AI’s strategic capabilities.

3. AI vs. ML: Bottom-Up vs. Top-Down

   Machine Learning (ML) builds intelligence from the bottom up, using statistics, logic, and optimization. In contrast, traditional AI aimed for a top-down approach — trying to directly simulate the human brain’s reasoning.

4. IBM & Backgammon (1994)

   One of the first major ML gaming successes was IBM’s backgammon AI. It used a neural network trained by self-play, continuously improving by playing millions of games against itself — and eventually beating the world champion.

5. Chess Breakthrough

   Building on its backgammon success, IBM applied similar ML techniques to chess — achieving world-class performance and paving the way for today’s AI milestones like AlphaGo.

📚 Machine Learning (ML) Fundamentals

1. Types of ML
   • Supervised Learning: Data comes with labels (answers).
   • Unsupervised Learning: Data without labels — the algorithm discovers patterns on its own.
   • Reinforcement Learning: Learning by trial and error — the model gets rewards or penalties for its actions to optimize its behavior.
2. Targets (Y)
   • Binary Classification: Classify inputs into two categories.
   • Multi-Class Classification: More than two categories.
   • Regression: Predict a continuous numerical value.
3. Inputs (X)
   • Known as features — the input variables the model uses to learn and make predictions.

✨ Key Idea:

AI and ML have evolved from handcrafted rules and logic to powerful models that learn from data. From checkers and chess to today’s deep neural networks, the journey shows how data-driven learning keeps pushing the limits of what machines can do.