Real quick, on this as well, I don't think the endpoint is that we will constantly have conversations with our phone.

I think that's actually, for them, that's just a necessary step along the way to the actual endpoint, is that conversation becomes obsolete because the AI can so precisely predict exactly what you need before you even think to ask for it,

you no longer have to directly interact with the phone.

It becomes a one-way interaction of the AI feeding you.

Oh, you want to know what the weather is?

I knew you wanted to know what the weather is. Before you even knew you wanted to know what the weather is? I knew you wanted to know what the weather

is. Before you even knew you wanted to know what the weather is, here's what the weather is, right?

That is, I think, actually they're thinking here. This is why, you know, all these models are

predictive models to varying degrees, whether it's predicting what the next token in a linguistic

chain is, or it's predicting what somebody

wants out of a search result, or it's predicting what you want before you know that you want

it.

That's their actual mindset here.

It's not that it becomes this like constant companion that you're chattering away with all

day, every day. Instead it becomes this thing that're chattering away with all day, every day. Instead, it becomes this thing that

makes chattering away obsolete because it's feeding you information that you then just receive.

So there are a couple other sides of that. Number one, I think the other side of the multimodal

model is, yes, there is the one that individuals will use as consumers.

I take your point, Jathan, as well, that like, and this is especially true if you look at

the other companies, Microsoft and Google especially, who's all their new product announcements

are just like, let us intrude in more of your life. Give us access to everything and we'll

let you search your own photos or whatever.

...