Was ich aus lokaler KI rausgeholt habe

If you're using AI, this is probably how it often works these days. You type something into a chat window and somewhere in a data center a model calculates for you. At some point your quota for today will be used up, unless you add more on top, in which case it can go up to subscriptions costing 500 euros

per month. OpenAI announced this just a few days ago. At the same time, a relatively large number of people are wondering whether the AI ​​bubble might burst soon. And so we occasionally ask ourselves , what does that actually mean for us? Today I want to show you what you can already do with your very own PC

, without having to touch any cloud model at all. This woman's name is Mira. The image model on the computer was completely unknown to her at first, until I taught it to her on her own graphics card. Meanwhile, I have a voice for her, a few colleagues, including a rather tired ficus (which is a tree), and I've

built my own little game around it. A large part, in fact the largest part, of the code, the images and the voices for them were created locally on this computer. And what came out in the end, I think is incredibly good. That's enough for 99 percent of practically everything. Alternate provided me with the computer for this today

. AMD and Samsung are also on board. And this entire video is a collaboration with all three of them. But I chose exactly what I'm going to build myself . Inside the computer is a Ryzen 7 9700X, a Radeon AI Pro R 9700 with 32 GB of graphics memory, 64 GB of RAM and a Samsung 9100

Pro SSD. At Alternate, you can easily configure a computer exactly how you want it. With exactly the components that suit you best. The AMD card makes a lot of sense, of course, because 32 GB of graphics memory allows for a lot of local processing . And what exactly? I'll show you in a moment. Local, however, means quite