যেকোনো AI স্ট্যাকের ৬টা লেয়ার

Hello everyone. Today I will talk about the six layers that every AI tool in the world works in. You might already know about a website called Futurepedia , where you can find about 54,000 to 55,000 AI tools. Now, it's not possible for a single person to learn all those tools. But just saying it's not

possible doesn't solve it. Right? We feel like we are falling behind. We feel like we are missing out. My colleagues around me know five tools. I only know two of those five. I don't know the rest. And while trying to learn those, we find 500 more new tools . So, this problem will persist as long

as we keep trying to learn new tools. It cannot be solved that way. Okay, so to solve this, what I have tried is that all these 54,000 or 55,000 tools, plus any tools that come in the future —I hope, unless a new category of tools is invented—like, you might know, and I'm saying this off-topic, a

new type of model has arrived, which is a System 1 model. It's not a normal Large Language Model, okay? I mean, it is an AI model, but not a Large Language Model. It cannot generate anything. Its name is, let me see the type, this one, it's not a generative AI. This is an AI for making decisions.

Decision—this is a probabilistic model that will actually tell you, very quickly and at a very low cost, the decision of exactly which one has the highest chance of being correct. So my main point is that, unless a new category of tools is invented like this , in that case, no matter how many

tools come out, we can put every tool into six boxes. You just need to know which tool falls into which box. The first three boxes are Chat, Reasoning, Generate, Retrieve, or these three boxes. The job of these first three boxes is to create something. Ultimately, the job of these three boxes, or layers—we're calling them