Reinforcement Learning is The Theory of Everything for AI—This is Your Guide to What it Is and How it Works

You get a car! You get a car! You all get a car! Remember that video from Oprah?

Thanks Oprah

Oprah’s reward giving is the simplest explanation for Reinforcement Learning I can think of—basically a mathematical theory of how reward availability shapes agent behavior over time. If everybody gets a car, how does that shape behavior over time? What if we only get a car in reward for certain behaviors? (Yes I know I’m stretching daytime television here.)

Don’t worry, even if you don’t watch Oprah I’ll explain the whole concept more simply down below, and explain why it matters and how it actually applies in specific businesses today.

I’m really excited about this piece. I think Reinforcement Learning is a powerfully simple concept when explained well, so I want to spend a lot of time in this post basically explaining what Reinforcement Learning is in detail, where it came from, and I want you to walk away with confidence understanding the real dynamics driving the AI revolution. Enjoy!