AI IS EVERYWHERE NOW

It's been more than a year since I started learning about AI... and honestly it's getting crazier day by day.

Back in early 2023 I was just experimenting with some AI tools out of curiosity. At that time I didn't really understand what was happening behind the scenes. It was just surprising to see a machine answering questions like a human. But over the past year things changed a lot.

Instead of just using these tools, I started trying to understand how they actually work.

I started reading about LLMs (Large Language Models), how they are trained, how massive datasets are used, and how models generate tokens step by step to produce responses. At first it felt very confusing. Terms like transformers, embeddings, context windows, and training parameters were everywhere. But slowly things started making a little more sense. two resources that helped me a lot were the paper Attention Is All You Need and the OpenAI API docs.

I also started exploring things like RAG systems (Retrieval Augmented Generation). The idea that an AI model can connect to external knowledge sources and retrieve information before generating an answer is really interesting. It feels like giving models a dynamic memory instead of relying only on what they learned during training. the original RAG paper gave me a clearer technical picture.

Another thing I found fascinating is memory systems for AI. Some developers are experimenting with ways for AI agents to remember previous conversations, store information, and use that context later when performing tasks. If these systems improve, AI could move beyond simple chat responses and start behaving more like long-running digital processes.

At the same time the progress in the AI ecosystem has been incredibly fast.

Companies like OpenAI, Anthropic, Google DeepMind, and Meta are building extremely powerful models. Systems like GPT-4, Claude, Gemini, and the LLaMA family are pushing the limits of what language models can do. Meanwhile open-source communities are building tools that allow developers around the world to experiment with these models in different ways.

But what interests me the most is not just the models.

It's what people are starting to build on top of them.

Developers are experimenting with AI agents, systems that can plan tasks, access tools, search the internet, write code, and complete workflows. Instead of a single prompt and response, these systems can run sequences of actions and interact with different tools.

That idea is extremely powerful.

Because when you start thinking about it deeply, many types of work in the world are basically structured processes. Research, planning, content creation, analysis, coordination -- many of these tasks follow patterns.

So I keep thinking about something.

What if instead of building isolated AI tools, we start building structured AI systems designed to perform specific roles?

Not just assistants.

But specialized digital systems that can research information, generate strategies, execute tasks, and coordinate outputs continuously.

Almost like a layer of digital specialists.

This is still just a rough thought in my head right now. Mostly hypothetical thinking.

But another concept that caught my attention recently is decentralized computing networks. Some projects are experimenting with distributed intelligence where multiple machines contribute models and computing power into shared networks. Instead of a single company controlling everything, intelligence could become something coordinated across many participants.

Ideas like this are being explored in networks like Bittensor, where machine learning models compete and collaborate inside decentralized systems. i used Bittensor docs as a starting point to understand this side.

Combining distributed computation with intelligent models could open completely new possibilities.

Right now I'm still very early in this journey. Mostly reading, learning, and experimenting.

But one thing feels very clear.

AI is not just another software trend.

It's becoming a foundational technology.

And looking at how fast things are moving, I have a strong feeling that within the next two or three years I want to seriously work on something in this space. Maybe start researching deeper into how intelligent systems coordinate work, how models interact with tools, and how distributed networks could support large scale AI infrastructure.

For now these are just ideas.

But the more I learn, the more it feels like something very big is coming.

And I want to be part of building it.