So now we know what the problem is, how do we fix it, right? Like what is happening now that is a different approach to fixing this problem?

The thing that is happening now that is tracking large language models and natural language processing is that we are trying to find the task that is the superset of every other task.

And natural language processing some 10 years ago was in a similar situation.

So natural language processing had all of these intermediate tasks, like given a sentence, can we classify the sentiment of the sentence as positive, negative, sad, happy, whatever.

Then maybe there was a thing called like grammars and parsing trees, where you basically take a sentence and then you're trying to say this is the verb, this is the object, subject, etc., etc., And as an application, people then said, oh, when we have solved all of these intermediate problems, then we can build a system that enables us to do dialogue.

Then we can talk to the system.

So, of course, history went differently.

What are the applications that are going to affect me?