All days
Day 01 DoneOctober 2025

Day 1: AI is literally a baby

Introduction to AI and Large Language Models — how models learn, and how the input/output relationship grew from text-only chatbots into multimodal systems.

AI is literally a baby!


A baby only knows as much as you can teach it. The same goes for AI; you have to carefully and intentionally allow it to learn to process the world and make its own connections.


We get really excited by how well the chatbots seem to solve our problems, but we never seem to understand how that's possible. The foundation of AI is built on the model's ability to understand what you tell it and what that input can produce.


ChatGPT started as just a chatbot that could take in text and spit out more text, but this limitation has been removed as the technology has progressed. We can now feed AI with photos, videos, and even voice recordings, and it's able to give us a response that can be any kind of media.


From calculators to models


This input-output relationship is a result of years-long teaching of computers about how the human world operates. How a "seed" (input) can produce a tree (an output). Basic computer programming has evolved from the first calculators in the 17th century to these highly intelligent models that can seemingly solve very complex math equations in an instant.


This is only possible because we learnt how to break down the complex world around us into a simple math problem that can then grow on its own. Such a superpower is reserved for engineers and programmers who only dare to imagine and think.


Thank you for reading. Let me know what you would love to learn.