These are the compute elements or the components that actually perform the computation.

that actually perform the computation.

So this cost savings along with the basically lightweight of the model, meaning you can download it and run it on a laptop, has sent kind of shivers through the spines because what this really means is there's a possible future where algorithmic efficiency beats brute force computation.

And the entire AI industry in the United States and around the world has been thinking about high-level computation and big data centers and a lot of energy use and chips.

So notice, we now have three areas of artificial intelligence development where a great deal of computation would result in higher quality answers.

And so I think that we're several years away from being able to do that, and in the meantime, we have to keep increasing our computation.

a little bit of computation, and then it dies. And I'm pressing

will be the beneficiary of these redesigns of how we do computation. It also helps that

TrueNorth and Loihi also combined into one chip two aspects of computers that are normally separate: memory and computation.

In a typical computer like you have at home, the CPU handles computation and shuffles data back and forth from the Random Access Memory, or RAM.

Given the huge number of parameters and the enormous amount of training data, the scale of computation involved in training a large language model is mind-boggling.

Looking back at the pre-training though, this staggering amount of computation is only made possible by using special computer chips that are optimized for running many, many operations in parallel, known as GPUs.

logic and computation, algorithms, and then operating systems and computer organization and design.

on AI computation made up less than half

I'll have to use all the computation power

The amount of computation we have to do in AI is so much greater as a result of reasoning AI and the training of reasoning AI systems and agentic systems.