🎨 GAN Explained Like You're 5
Building AI systems and writing about how they actually work. Master of AI @ University of Technology Sydney. Previously B.Tech CS with focus on IoT. I believe the best way to learn is to explain. That's why I'm documenting tech concepts with simple analogies (@sreekarreddy.com). AWS Certified • Azure AI Certified • Neo4j Professional • Google Data Analytics When not coding: exploring Sydney, working on side projects, and teaching tech to anyone who'll listen.
Two networks competing to create realistic content
Day 76 of 149
👉 Full deep-dive with code examples
The Forger vs Detective Analogy
Imagine an art class with:
The Forger (Generator):
- Tries to create fake Picasso paintings
- Wants to fool the expert
The Detective (Discriminator):
- Tries to spot the fakes
- Wants to correctly identify real vs fake
They compete:
- Forger gets better at fooling
- Detective gets better at spotting
- Both improve!
Eventually, fakes are indistinguishable from real!
How GANs Work
Random Noise → [Generator] → Fake Image
↓
Real Images ─────→ [Discriminator] → "Real" or "Fake"?
↓
Both learn from mistakes
The generator learns to create increasingly realistic content.
What GANs Can Create
- Faces: thispersondoesnotexist.com (none are real people!)
- Art: New paintings in any style
- Photos: High-resolution image enhancement
- Deep fakes: Face-swapped videos
The Dark Side
GANs power deep fakes - fake videos of real people.
Used for:
- Entertainment (face swaps)
- Misinformation (fake political videos)
- Fraud (identity theft)
Why GANs Are Impressive
They CREATE new content, not just classify.
Before: "Is this a cat?" (classification) Now: "Make me a new cat!" (generation)
In One Sentence
GANs pit two neural networks against each other - one creates fakes, one detects them - until the fakes become indistinguishable from reality.
🔗 Enjoying these? Follow for daily ELI5 explanations!
Making complex tech concepts simple, one day at a time.