The Origins of the Animals Program: A Digital Guessing Game with a Brain

Ah, the Animals program — a humble, text-based guessing game that turned many early computer users into amateur AI enthusiasts.

If you had an Apple ][ back in the day, you might remember this quirky little program that seemed to learn from your mistakes (just like a stubborn dog figuring out how to sneak snacks).

While Eliza was busy playing psychotherapist, Animals was quietly demonstrating how computers could actually learn. It was simple, clever, and a sneak peek into the world of expert systems and machine learning.

How Did It Work? Magic? Nope. Decision Trees.

The Animals program was like a digital game of 20 Questions, only without the smug attitude. It followed a straightforward process:

1. Start with a basic tree: The game began with just a couple of animals and yes/no questions.
2. Ask yes/no questions: It tried to guess your animal by narrowing down possibilities.
3. Learn from failure: If it guessed wrong, it asked you to provide a new question to distinguish the new animal from the old ones.
4. Save the new knowledge: The next time you played, it used your new question to guess better.

Here’s a sample session:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

**Think of an animal. I will try to guess it.**
> OK
**Is it a mammal?**
> Yes
**Does it have a long neck?**
> No
**I give up! What was your animal?**
> Kangaroo
**What question could I have asked to distinguish a kangaroo?**
> Does it hop?
**Thank you! I will remember that.**

And just like that, the computer’s mental animal zoo expanded.

The Learning Mechanism: A Basic Expert System

The Animals program wasn’t just guessing blindly; it was building an expert system — a program that uses a knowledge base of facts and rules to make decisions. Each new animal added a branch to its decision tree.

• Expert Systems: Rule-based systems that apply logic and facts to solve problems.
• Decision Trees: Like a flowchart for animal guesses.
• Rule-Based Systems: The “if-then” logic behind the game. For example: If it’s a mammal and it hops, it’s a kangaroo.

The Connection to Neural Networks

So how does the Animals program relate to modern AI? Well, it’s like comparing a tricycle to a Tesla:

• Animals: Rule-based, static knowledge, requires human input to learn.
• Neural Networks: Data-driven, dynamic learning, no hand-crafted rules.

Yet, the program laid the conceptual foundation for understanding machine learning principles:

• Learning from experience: Animals stored new knowledge.
• Pattern recognition: It categorized animals based on shared characteristics.
• Interactive feedback loop: User input helped it improve, much like supervised learning.

My First Encounter with Animals

I first stumbled across the Animals program on my Apple ][ when I was supposed to be doing homework. The game asked me if my animal was a mammal, and when it confidently guessed “aardvark” for my kangaroo, I felt a rush of power as I taught it the “Does it hop?” question.

Little did I know, I was participating in a rudimentary form of AI training.

Animals vs. Modern AI: The Evolution

Key Differences:

• Hardcoded Rules vs. Self-Learning Algorithms: Animals needed me to add rules; modern AI learns patterns from data.
• Small Knowledge Base vs. Big Data: Animals remembered a handful of animals; Modern AI Bots know more trivia than your local pub quiz champion.
• Simple Decision Trees vs. Deep Neural Networks: One branched Q&A game vs. multi-layered algorithms.

Shared DNA:

Both use questions to categorize information and improve performance with experience.

Legacy of the Animals Program

The Animals program may seem laughably simple today, but it introduced countless users to machine learning concepts.

It showed that computers could learn from their mistakes — a revelation in the late 1970s.

Key Ideas

• Animals was an early AI-like program on the Apple ][.
• It used decision trees and rule-based logic.
• It demonstrated machine learning principles to everyday users.
• Its structure resembles the foundation of expert systems.
• It paved the way for understanding neural networks and modern AI.

References

1. https://en.wikipedia.org/wiki/Expert_system
2. https://liacademy.co.uk/the-story-of-eliza-the-ai-that-fooled-the-world/
3. https://builtin.com/artificial-intelligence/eliza-effect
4. Ford, M. (2015). Rise of the Robots: Technology and the Threat of a Jobless Future.
5. Russell, S. & Norvig, P. (2010). Artificial Intelligence: A Modern Approach.