This is CBSI on the world. I'm John Batchel, spending time with Gary Rivlin. His new book is

AI Valley. Revolatory. If you puzzle about AI or you've poked around in it or you're holding back,

there's no reason. These are the same people who brought you the social network we spend our days on. These are the same people who brought you the social network we spend our days on.

These are the same people who brought you the genius of LinkedIn and other very helpful programs

that we can't remember how we live without them.

All right.

Now we go to an important development.

I have it down to 2017, but it might be a different year.

The Transformers paper about large language models.

What is that, Gary?

How does it contribute?

Right.

So a group of researchers inside of Google came up with the Transformer paper.

And, you know, just essentially it's just like a human doesn't necessarily read word by word.

You understand things in context.

And so the transformer paper was a way of allowing computers to divine a meaning from the context

in which the word, the phrase was being used.

And that would be essential.

It would be essential to kind of the chatbots that were all used to.

Before that, they weren't really having very much luck getting computers to talk in natural

language in English or whatever language they were being taught in.

But the Transformer paper really kind of help folks or help the people developing AI create chatbots that could talk to us in normal language.

...