Skip to content

4. Train a Computer Wrap-Up

Playback Speed:
Transcript

Now that you have trained a computer to identify items, think about how you recognize everything in the world around you.

For example, why does it take only a quick glance to know something is a table, a bird, or a street sign?

Cats and dogs are about the same size.

They both walk on four legs and have fur, tails, and whiskers.

So how can you tell the difference between them almost immediately?

In general, humans are excellent at item identification.

Our brains can categorize and organize data very effectively and put it into context.

But sometimes, it is more difficult.

Now, imagine you are a computer trying to differentiate a cat and a dog.

Computers do not process information the same way humans do, and they cannot put things into context.

Instead, they translate words and images into binary code -- a series of ones and zeroes -- and then process that code to get a result.

This makes it very difficult for a computer to do tasks that most people take for granted.

For example: Handwriting analysis and voice-to-text both require very complex models.

There are thousands of ways to write a word and countless different accents that change its pronunciation ... And an infinite number of ways to draw even a simple image.

So, the next time your social media app tags the wrong person or your phone does not translate your voice to text correctly, remember how difficult it is for computers to make those decisions!

Next