How AI Thinks: Neurons Demystified

Ready to Publish

Publish Date

Nov 13, 2023 23:00

Slug

How-AI-Thinks-Neurons-Demystified

Today’s One Thing I’m thinking about in my pursuit of better involves artificial neural networks and understanding the distinction between a neuron and a connection.

I’m studying machine learning on LinkedIn as part of a basic Career Essentials in Generative AI by Microsoft and LinkedIn Certification.

AI-generated graphic illustrating the comparison between neurons and connections in an artificial neural network. This image visually contrasts the organic complexity of biological neurons on the left with the geometric simplicity of an artificial neural network on the right. —ChatGPT 4 Turbo

Artificial Neural Networks are a type of Machine Learning (ML) that loosely mimic the structure of the human brain. Depending on the task's complexity, these networks range from dozens to millions of neurons. Each neuron acts as a processing unit that receives input, processes it using an activation function, and produces an output. The neurons are interconnected through connections, and each connection has a weight that modulates the signal passing through it. Alongside these weights, each neuron has a bias that allows the activation function to be shifted, contributing to the network's ability to learn and make predictions.

AI-generated graphic showing a real-world analogy using a person as a neuron and others giving advice as a connection. —ChatGPT 4 Turbo

In a real-world analogy, we can think of a neuron as a person making a decision, where the connections are like the pieces of advice or information they receive from others. The 'weight' of the connection is analogous to how much trust the person places in each piece of advice. In artificial neural networks, 'bias' refers to a parameter that adjusts the neuron's output, akin to a person’s individual tendency or preference influencing their final decision.

One Thing to Better: Reflect on how this analogy of neurons and connections in AI can apply to our daily decision-making. Just as a neural network processes information to reach a conclusion, consider what 'weights' you assign to the advice and information you receive and how they guide your decisions.

Until next time, choose your One Thing and forge your unique path to better.