Tokens
What a token is, why Hindi uses more tokens than English, the strawberry bug, and why limits and cost are counted in tokens.
- AI Fundamentals
- Beginner
- 3 min read
- YouTube video
AI Explained series
One AI concept per episode, with something you can try today. Each lesson has the full explanation, the video links, sources and a quick quiz.
What a token is, why Hindi uses more tokens than English, the strawberry bug, and why limits and cost are counted in tokens.
Why a long ChatGPT chat forgets your first messages: the context window is AI's desk. Plus 3 tricks to stop it forgetting.
Why AI gives confident wrong answers, fake links and fake facts, the $100 billion mistake, and 3 tricks to catch it.
RAG (Retrieval Augmented Generation) as an open-book exam for AI, in 4 steps, how it differs from fine-tuning, and how to try it free in NotebookLM.
Stop typing "resume bana do". Add 4 things to any prompt (Role, Context, Task, Format) plus 2 bonus tricks. With the free PROMPT sheet.
Planned topics. No article or video yet.
Imagine AI researches a topic, organises what it finds and prepares a report on its own. How does it do that? Tools and multi-step tasks, explained.
Promised at the end of Episode 5. Next up.
How AI turns meaning into numbers, and why that lets it find similar ideas instead of matching keywords.
Training, next-token prediction and what an LLM can and cannot do, tied together in one lesson.
Where RAG is useful day to day: chatting with your notes, policies and manuals, and what to check before trusting the answers.
The hidden instructions an app gives AI before your message, and how they differ from what you type.
Why the same prompt can give a safe answer one time and a creative one the next, and what the temperature setting changes.
No episode matches that search.