Benjamin Lee

So I don't see how you can, under current infrastructure investment plans, you could possibly achieve those net zero goals.

And as you can imagine, it's much more efficient because now you're just cooling the surface of whatever the cold plate is covering rather than just blowing air through the entire machine.

The whole idea behind a mixture of experts is you don't need a single huge model with a trillion parameters to answer every possible question under the sun. Rather, you would like to have a collection of experts. smaller models, and then you just sort of route the request to the right expert. And because each expert is so much smaller, it's going to cost less energy to invoke.

Before generative AI came along in late 2022, there was hope among these data center operators that they could go to net zero.

The first part is ingesting system. So the second part is the collector system.

Yeah, it's similar. The principle is the same for the vacuum cleaner. Yeah, that's right.

You push the cool air through the front of the machines, and all the warm air gets pushed out the back.

I think the difficulty has been... That the air conditioning infrastructure is having trouble keeping up with the latest in GPUs and how closely packed GPUs are.

They are delivering more performance, but they also may be drawing more power, which is why we are now taking unprecedented steps to cool them.

The search engine was a killer app. Another example of that would be a recommendation system that social media feeds use to recommend ads and content. That was a killer app. But we don't have that for generative AI.

There's a lot of generative AI being invoked on your behalf as these companies try to figure out what it's good for.

I think Before generative AI came along in the late 2022, there was hope among these data center operators that they could go to net zero.

I don't see how you can, under current infrastructure investment plans, you could possibly achieve those net zero goals.

For me, as a mental health professional, I get really cautious when using, even using the word trauma. No one actually comes in and says, hey, I'm dealing with intergenerational trauma. Can you help me? Right. You know, they don't have the language. People don't have the language for it. And so part of that. The generational legacy of Black families is we don't talk about our problems.

We just kind of roll through them. We deal with them. We're strong. And we just keep it moving forward.

A silence, it eats at our bodies. It really does. It's almost like every time we're silent, it's like a little pinprick that we do to our bodies internally. And after so long, those little pinpricks turn up as heart disease, as cancer, as, you know, all these other ailments.

Imagine every time you walk out of the door, your body, you're tightening your body, you're tensing up your body, and you're holding onto it for the entire day. until you come home at night. What do we think would happen to our bodies as a result?

That's what that hypervigilance does. That hypervigilance causes our bodies to tense up so that we can't fully breathe. Yeah.

You know, we're not at the point yet where we can make these definitive connections that, yes, this specific gene expression has been passed down. But, you know, we are passing down this idea of hush, right? Stay silent. Don't fight because you don't have power.

This generation. pain has compounded over generations and so we're going to have to deconstruct it or heal it over generations right you know our generation and the generations that come behind us will have little pieces of the work to do as we put mental health more so at the forefront. And as we start to communicate more and more within our families, that's how we engage in this healing.

Because now you're just cooling the surface of whatever the cold plate is covering rather than just blowing air through the entire machine.

The idea behind a mixture of experts is you don't need a single huge model with a trillion parameters to answer every possible question under the sun. Rather, you would like to have a collection of experts, smaller models, and then you just sort of route the request to the right expert. And because each expert is so much smaller, it's going to cost less energy to invoke.