# @itsemilyhiggins on Instagram

- **Type:** Video
- **Original URL:** https://www.instagram.com/p/DX12epYOWfN
- **Gondola URL:** https://gondola.cc/posts/65134224-itsemilyhiggins-instagram
- **Thumbnail:** https://img.gondola.cc/tr:w-,h-,fo-auto/postThumbnails/f00e1b6a2b.jpg
- **Posted:** 2026-05-02T15:46:47.000+00:00
- **Account Owner:** Emily Higgins (@itsemilyhiggins) — https://gondola.cc/itsemilyhiggins

## Caption

Welcome to episode 1 of The Thinking Machine: my new series that compares real and artificial intelligence 🧠👾

You don’t see reality. You see your brain’s prediction of it.

Right now, your brain is generating a constant stream of guesses about what’s about to happen. 

The next word in this sentence. 
The temperature of the air on your skin. 
The weight of your phone in your hand. 

It compares those guesses to what your senses actually report, and most of the time, the prediction is close enough that you don’t notice the process is happening at all. Your experience of “reality” is mostly just successful predictions.

When a prediction is wrong, something more interesting happens. The mismatch is called prediction error, and it’s how you learn. Your brain updates the model so the next guess is better. 

Trip on a step that was shorter than expected, and your motor system quietly recalibrates. 
Hear a word you didn’t expect, and your language model updates. 
This loop — predict, compare, update — is running every second of your life. 

Neuroscientists call the framework predictive processing, and a lot of them now think it’s the closest thing we have to a unifying theory of how the brain works.

ChatGPT runs on a structurally similar loop. It generates text by predicting the next token (a token is roughly a word or piece of a word) based on everything that came before. But the more interesting parallel is how it got good at that in the first place. During training, the model was given massive amounts of text with words hidden, asked to predict the missing ones, and then shown the right answer. Every wrong guess produced an error signal, which got fed back through the network to adjust billions of internal parameters, slightly. Do that trillions of times and you end up with a system that’s eerily good at predicting what comes next. 

Sound familiar?

So the structural overlap goes deeper than “both do prediction.” Both systems learn by being wrong. Prediction error is the signal that drives the update, in the brain and in the model.

#aieducation #psychology

## Stats

- **Views:** 6,444
- **Likes:** 476
- **Shares:** 0
- **Comments:** 5

## Tags

aieducation, psychology

---
Copyright (c) Gondola