ChatGPT Library

Recommended Links

ChatGPT Glossary

ChatGPT

An AI chatbot built by OpenAI that can answer questions, write, code, and chat in natural language.

Analogy: Like texting a super-knowledgeable friend who never sleeps.

GPT (Generative Pre-trained Transformer)

The type of AI model powering ChatGPT. 'Generative' means it creates text; 'pre-trained' means it was taught on lots of data; 'transformer' is the tech behind it.

Analogy: Think of GPT as the 'engine' under the hood of ChatGPT.

OpenAI

The company that created ChatGPT, plus other AI like DALL·E (images) and Sora (video).

Analogy: If AI were Marvel superheroes, OpenAI would be the studio making the films.

Prompt

The text or question you type into ChatGPT.

Analogy: Like giving an instruction to a personal assistant.

Context Window

The amount of conversation ChatGPT can 'remember' at once.

Analogy: Like its working memory or notepad—too much, and it forgets the beginning.

Hallucination

When ChatGPT makes up information that sounds real but isn’t.

Analogy: Like a confident friend who sometimes lies by accident.

Model

A specific version of ChatGPT’s 'brain' (e.g., GPT-4, GPT-5).

Analogy: Think of it like iPhone versions—newer models usually have more features.

Training Data

The text, code, and examples used to teach ChatGPT how to generate language.

Analogy: Like the books, articles, and notes you studied for an exam.

Machine Learning

A way computers improve by finding patterns in data.

Analogy: Like learning to play piano by practicing songs over and over.

Transformer

The AI architecture that makes GPT models so powerful at language.

Analogy: Like the assembly line in a factory that processes words (tokens) at lightning speed.

Custom GPT

A personalized ChatGPT you can create with custom instructions or tools.

Analogy: Like programming your own digital intern.

Guardrails / Safety Filters

Systems that stop ChatGPT from giving harmful or unsafe outputs.

Analogy: Like seatbelts and airbags for your conversations.

Bias

When AI reflects imbalances or stereotypes from its training data.

Analogy: Like a funhouse mirror—it reflects reality, but sometimes in a distorted way.

Token

A chunk of text (word or part of a word) that ChatGPT reads.

Analogy: Like LEGO bricks that build sentences.

FAQ’s

  • Access to ChatGPT is free, with limitations. Paid subscribers receive higher usage limits,

    priority access during peak times and access to other models.

  • Writing, summarizing, translating, coding, tutoring, brainstorming, technical support, meeting prep, presentations, and more.

  • These models were trained on vast amounts of data from the internet written by humans, including conversations, so the responses it provides may sound human-like. Outputs may be inaccurate, untruthful, and otherwise misleading at times.

  • Your chats aren’t used to train OpenAI models if you opt out in settings. Business/enterprise accounts have enhanced data protections.

  • It does occasionally “hallucinate” (i.e., generate false information).

    Always double-check facts. Think of it like a super-intelligent intern — fast, but not always right.

  • Yes! It’s available on iOS and Android and has a clean interface, voice chat, file uploads, and Custom GPT support.

  • Yes, it can read and analyze most formats, including PDFs, Word, Excel, and images.