What is Human feedback?

Human feedback is the process of people giving hints and corrections to help computers learn how to do things better.

Imagine a robot learning to catch a ball. At first, it guesses wildly. Then, you say, "Too high!" or "Nice throw!", and it adjusts its brain until it gets it right every time. That simple exchange is the core of human feedback.

Computers start out like babies who don't know what anything is. They see a picture but can't tell if it is a cat or a toaster. Humans step in to show them the difference.

1. The Teacher: A person looks at data and marks the correct answer, just like checking homework with a red pen.
2. The Guide: Instead of just saying "wrong," we say "this one is better than that one." It is like choosing your favorite cookie out of two bags. The computer learns to prefer the chocolate chip over the plain oatmeal because you picked it twice as often.

We use human feedback because computers are very good at math but not so good at common sense. They might write a perfect poem but forget that dogs have four legs, not two. Humans provide the real-world context that data alone cannot offer. When we correct errors, we teach the machine what matters to us in our daily lives, turning raw numbers into useful helpers.