Grant

So there are those use cases that just make common sense, makes perfect sense.

But let's say you want to be a software engineer and you're now offloading all of your thinking to Codex or Cloud Code and you're not learning anything that you're doing on the back end.

There's a way to do it where you are, where you basically send a request, then you go read the code, then you ask questions about the code.

What is this?

What does this function do?

Then you read it yourself and try to say, okay, I understand, you know, I can see here to here, it does this and that, and you're wrestling with it, but that still takes time.

That's not saving you the time that an agent would, when you could just ask it to build you software and it builds you software.

The demo that they gave on Monday was just like incredible how

goodness thing is getting.

So the question is, well, on an individual level, do I personally want to learn this or not?

Or do I just personally want to have done this?

But then when you apply that at the industry wide level, like the incentive, all the incentives are there for them to create tools to basically do

everything for us because they can sell it to companies.

And then on a society-wide level, well, we want the GDP growth.

Well, we want this and that.

We want to improve technology.

And we're in a global race to create AGI, whether or not you've subscribed to whether we should or shouldn't do that.

So there's all of these incentives that is pushing us to automate as much as possible.

So maybe the companies are going to try and automate as much as possible.

Maybe governments are going to try and automate as much as possible.