Freeberg

But the unfortunate reality in most AI workloads is that you end up with something that's not very good. And that's not the fault of OpenAI, nor is it the fault of Lama. It's just the nature of how these bottles work and the fact that, you know, you've gone from something very, you know, deterministic to something that's very probabilistic.

But the unfortunate reality in most AI workloads is that you end up with something that's not very good. And that's not the fault of OpenAI, nor is it the fault of Lama. It's just the nature of how these bottles work and the fact that, you know, you've gone from something very, you know, deterministic to something that's very probabilistic.

And we have not found a way to create extremely high quality experiences, we as an industry. that make this cost justifiable. So right now, as you said, Jason, most of that is toy apps and a lot of experimentation. Will we get there? I absolutely think so. But we're not there yet. So I suspect that most of the revenue then is on the consumer side right now, actually.

And we have not found a way to create extremely high quality experiences, we as an industry. that make this cost justifiable. So right now, as you said, Jason, most of that is toy apps and a lot of experimentation. Will we get there? I absolutely think so. But we're not there yet. So I suspect that most of the revenue then is on the consumer side right now, actually.

And so I just looked at Google Trends, and I first looked at the United States, and I just wanted to understand how is the traction. And what's interesting is, is that

And so I just looked at Google Trends, and I first looked at the United States, and I just wanted to understand how is the traction. And what's interesting is, is that

The traffic in the United States is down about 25%, 26% since the end of the school year, which I think reinforces this belief that ChatGPT has definitely over-indexed in that young people's cohort where they rely on it to write essays or what have you. But then I looked outside the United States and worldwide. And what's interesting about that is that's actually still growing up into the right.

The traffic in the United States is down about 25%, 26% since the end of the school year, which I think reinforces this belief that ChatGPT has definitely over-indexed in that young people's cohort where they rely on it to write essays or what have you. But then I looked outside the United States and worldwide. And what's interesting about that is that's actually still growing up into the right.

So I think you have then an ARPU problem, which is the same thing that we had to deal with at Facebook, which is we have these extremely valuable users in America. They asymptote at some point, and then we have to grow in other markets, but those markets aren't nearly as lucrative. Right. Yeah. They just don't generate nearly as much revenue.

So I think you have then an ARPU problem, which is the same thing that we had to deal with at Facebook, which is we have these extremely valuable users in America. They asymptote at some point, and then we have to grow in other markets, but those markets aren't nearly as lucrative. Right. Yeah. They just don't generate nearly as much revenue.

So then, I don't know, I would just wrap up by saying that I think that it's just, if this number is right, it's just an incredible testament to what they've been able to uncover and create in a short amount of time. That is legendary stuff. Yep.

So then, I don't know, I would just wrap up by saying that I think that it's just, if this number is right, it's just an incredible testament to what they've been able to uncover and create in a short amount of time. That is legendary stuff. Yep.

But what I can tell you, in working with enterprises around these AI experiences through 8090, we have not yet seen production quality models being deployed in the wild that are dealing with very, very important business processes, which means that most of the revenue right now is probably consumer oriented.

But what I can tell you, in working with enterprises around these AI experiences through 8090, we have not yet seen production quality models being deployed in the wild that are dealing with very, very important business processes, which means that most of the revenue right now is probably consumer oriented.

And were you paying them an enterprise license or a consumer license?

And were you paying them an enterprise license or a consumer license?

No, no, I was asking like the free bird. Yeah, no, free bird. Like you're just paying it yourself or you pay for everybody at Koala?

No, no, I was asking like the free bird. Yeah, no, free bird. Like you're just paying it yourself or you pay for everybody at Koala?

I will put myself on the other side of that trade if there was one. If I had to simplify the United States economy, remember, GDP is 70% spending by individual people. And so people can spend two things, savings that they have or credit that they have.

I will put myself on the other side of that trade if there was one. If I had to simplify the United States economy, remember, GDP is 70% spending by individual people. And so people can spend two things, savings that they have or credit that they have.