Skip to content

ChatGPT for Humanities Scholars

March 2026

Don't Panic

Skip to what you need: If you only use ChatGPT's conversational interface, read the orientation below and Chat, then skip to the Essentials section. If you want ChatGPT to do structured web research, see Research, Apps & Agent. If you are doing genuine coding work, see GPTs & Codex.

What ChatGPT is — and how it relates to OpenAI

OpenAI is an AI research and deployment company. It was founded in 2015 as a nonprofit, and its stated mission is to ensure that artificial general intelligence benefits all of humanity. In October 2025, OpenAI restructured. The nonprofit became the OpenAI Foundation, and the for-profit subsidiary became OpenAI Group, a public benefit corporation (PBC). The Foundation retains governance control over the for-profit entity. OpenAI develops models, consumer products, business products, and developer tools. ChatGPT is OpenAI's main general-purpose chat application: the product most people use directly in a browser, desktop app, or mobile app.

A very short history helps. ChatGPT launched in November 2022 as a conversational interface for interacting with OpenAI's models. Since then it has grown from "chatbot" into a broader work environment: ordinary chat, web search, file uploads, data analysis, projects, memory, voice, image generation, custom GPTs, deep research, apps that connect to third-party services, and more recently agentic task completion.

It is worth separating ChatGPT from the OpenAI API. ChatGPT is the end-user application: you open it and work inside it. The API is the developer platform for building your own tools, automations, websites, or research workflows on top of OpenAI models. For most humanities researchers, ChatGPT is the place to start. The API only becomes necessary when you want structured automation, reproducible pipelines, or integration into your own software. OpenAI bills the API separately from ChatGPT subscriptions.

Who this guide is for

This section is for humanities scholars who want practical guidance on using ChatGPT well. It is not a general introduction to AI. If you are new to these tools altogether, start with the Primer first. If you already know the basics and want to know which parts of ChatGPT matter for real research, teaching, writing, and administration, start here.

For prompting principles that apply across all platforms, see Prompting Principles. For data governance, verification, multi-model strategies, and cost awareness, see the Essentials section.

Which ChatGPT tool do I need?

ChatGPT is no longer just a chatbot. It offers several tools and modes, each suited to different kinds of work. Before going deep on any one, the guide below will help you choose where to start.

Chat

Best for: Asking questions, getting writing help, analysing documents, brainstorming.

You'll need: A free account or a paid subscription.

Skills required: None — just a browser.

Read the Chat guide →

Projects

Best for: Working on a research project with persistent context across sessions.

You'll need: Available on paid plans.

Skills required: None.

Read about Projects →

Deep Research

Best for: Structured web-based synthesis with a documented report.

You'll need: Available on Plus, Pro, and higher plans.

Skills required: None — but knowing what to ask for helps.

Read about Deep Research →

Apps & Agent

Best for: Connecting to third-party services or delegating complex multi-step online tasks.

You'll need: Available on paid plans; app availability varies.

Skills required: None for apps. Agent mode requires clear task specification.

Read about Apps & Agent →

Canvas

Best for: Iterative writing and editing alongside the model.

You'll need: Available on most plans.

Skills required: None.

Read about Canvas →

Custom GPTs

Best for: Repeated specialist tasks with a reusable configuration.

You'll need: Available on paid plans.

Skills required: None — but you need a clear, recurring need.

Read about GPTs →

Codex

Best for: Software engineering, building tools, scripts, pipelines.

You'll need: Available on paid plans (usage scales by tier).

Skills required: Genuine coding knowledge.

Read about Codex →

Many tasks start in Chat. Move to Projects when the work has continuity. Use Deep Research when you need broad web-based synthesis. Build a custom GPT only once a recurring need is clear. Ignore Codex unless you really do want software help.

The tool spectrum

Conversational Agentic
Chat Projects Deep Research Custom GPTs Agent / Codex
You talk, ChatGPT responds Persistent workspace Structured web synthesis Reusable configurations Delegated action

The ChatGPT tool spectrum: from conversation to autonomous action.

Decision flowchart

Question If No If Yes
Do you need structured web research? ↓ Deep Research
Is the work ongoing across sessions? ↓ Projects
Do you repeat the same kind of task often? ↓ Custom GPTs
Are you building software? Chat Codex

Which ChatGPT tool do I use? A quick decision guide.

Essential

If you mainly read, write, teach, and analyse texts: Start with Chat. Use Projects for sustained work. You can safely ignore Codex, agent mode, and custom GPTs until you have a specific need. The cross-platform essentials still apply to you — see Essentials.

If you want to build tools, transform data, or automate processing: Consider Codex or the OpenAI API. The Essentials still apply.

Minimum safe practice

If you just want to try ChatGPT responsibly:

  1. Use Chat. Upload the source text you want to work with.
  2. Ask for critique, summary, or analysis — not for authoritative facts.
  3. Verify all references against a library catalogue.
  4. Do not upload sensitive, personal, or embargoed material.
  5. Record what you asked and what you used from the response.
  6. If it's useful, great. If not, nothing was lost.

Contents