Microsoft is going deep in the AI game #artificialintelligence #ainews #chatgpt

So, I'll try to be quick about this. I don't have any good visuals, but Microsoft, the research team for AI, announced Cosmos 1, is what it's called, and it's a multimodal large-language model, so MLLM. If you don't know GPT, DeepMind, all that stuff, they're LLMs, large-language models. This is multimodal. What does that mean? Well, in education, we talk about learning a multimodal style of learning is How do I put this? It's all of those inputs, essentially. Vision, touch, you know, we learn by doing a lot, not just speaking. There's also seeing, and that's the thing with Cosmos, is Cosmos can learn by seeing. It took the RAVEN IQ test, which is based on nonverbal reasoning, and it did pretty good, and that's the thing here, is what we're getting towards is, imagine something powered by Cosmos, which, by the way, did it on less parameters. So, it's doing a lot more with less, which is very important here. But with that, you could do something like you do with chat GPT, but instead, you take a photo of, let's say, your sink, and it's leaking, and you take a photo, and go, why is this leaking? You don't even have to say that, actually. You could just take a photo, and they would see, hey, your seat's leaking because of this. Here are the steps to fix it, right? It can take in a visual process, and spit out an answer for you, based on that. So, it can do a lot more than just a text prompting. Very big news. They just launched a research paper. Things are moving so quickly, you guys, and this is just another example of that. Check it out, if you have a chance. It's an archive, and all over Reddit. Seriously, it's Cosmos with a K. K-O-S-M-O-S-1.