When you're friends with a software engineer

@devslopes

807.8K followers · usually 3.7K views a post

Views

40.3K

Likes

2.2K

Comments

31

Shares

174

SoundTikTok sound · original sound - devslopes

Why it worked

The video uses a relatable premise (having a software engineer friend) and a humorous, slightly inaccurate analogy to explain a complex topic (how LLMs work) in a simple and engaging way. The use of on-screen text and a conversational tone makes it easily digestible for a broad audience.

Summary

The video humorously explains that Large Language Models (LLMs) like ChatGPT are not sentient beings with their own brains. Instead, they are trained on vast amounts of internet data, allowing them to interpret human language.

Structure

1Introduction to having a software engineer friend
2Analogy of ChatGPT being its own human
3Correction: LLMs are trained on internet data
4Explanation of how LLMs interpret language
5Clarification: LLMs do not have their own brains

On-screen text

When you're friends with a software engineer yeah it's crazy it's like ChatGPT is its own human That's now how that works LLM's are trained on large sets of data gathered by the internet They're basically fed enough examples to where they can interpret human language they don't have their own brain

Original caption

When your friends a software engineer 🤓 #softwareengineer #csmajor #computerscience

More from @devslopes

More breakout videos from this creator.

Transcript

Yeah it's crazy. It's like ChatGPT is its own human. That's now how that works. LLMs are trained on large sets of data gathered by the internet. They're basically fed enough examples to where they can interpret human language. They don't have their own brain.