Hey everyone! I am Haren. Strangely curious and stupid guy. I am here to figure out why and how there’s AI. i know there’s no intelligence in AI as we know how it works in biological form but i am having hard time how the pure math can get us something that is close to or almost intelligent.
Very generally:
It’s all based on using a set of data (measured inputs and known-correct outputs) to create a mathematical model of the relationship between the inputs and the output.
A simplest example: Fitting a straight line to a bunch of measurements of distance vs. time, for a vehicle moving at a constant velocity. Time is the input, distance is the output.
The model will consist of a weight and a bias. In this example, they’re velocity and the initial position. You don’t need to know either, there is a mathematical process that “trains the model”, by computing the weight and bias that give the best fit.
Once you’re created the model, you can use it to make predictions for that system.
All of the implementations of AI are just variations of this theme, but using more and more complicated sets of data (like images, or the collected works of Shakespeare), which require more and more complicated models, which can make predictions like new images, or a text chat output.
1 Like
If I may add to Tmosh, it’s in large part due to the amount of data it’s trained on. These models have been trained almost all of the internet.
Have you ever wondered how is it that when you google the most random, unique thought that comes to your mind, search engine will find it? how it seems they have almost everything?
these models are simply predicting the next word based all of the information in the internet they can remember.
1 Like