chatgpt-images-documents-and-data

Images, Documents and Data | ITIAN Knowledge Hub

Module 4 · Lesson 3 of 4 · Course Lesson 14 of 26

Images, Documents and Data

Turn visual and file-based material into useful answers, summaries and charts. Learn which workflow fits each file, how to write a precise request and what you must verify before relying on the result.

Beginner
About 22 minutes
Interactive readiness check
4-question check

Learning Objectives

What you will learn

By the end of this lesson, you will be able to:

Distinguish image interpretation, document extraction and structured data analysis.
Prepare a file so ChatGPT can work with it more reliably.
Write a focused request that defines the task, context, output and checks.
Use follow-up questions to inspect evidence, calculations and uncertainty.
Recognise common problems with unclear images, scanned files, complex layouts and untidy spreadsheets.
Protect private, confidential and copyrighted material before uploading it.

Choose the Right Workflow

One upload button, three different jobs

ChatGPT handles visual images, text-rich documents and structured data differently. Start by identifying what the information actually is.

Images

Use visual interpretation for photos, screenshots, diagrams and charts.

Describe visible content
Read clear text
Compare visual options
Suggest edits or improvements

Documents

Use text extraction for PDFs, Word files, presentations and plain text.

Summarise sections
Find named information
Compare documents
Create actions or study notes

Structured data

Use code-backed analysis for spreadsheets, CSV files and other structured formats.

Clean and group rows
Calculate totals and trends
Find outliers
Create tables and charts

Availability and limits vary

Supported tools, file types and upload allowances can vary by model, plan, workspace settings and account capability. If an option is missing, check your current plan and the official file-upload guidance.

Practical Uses

Useful tasks for each input

Ask about a photograph

Request a neutral description, composition feedback, visible accessibility issues or ideas for a caption. Mark the area you want examined when precision matters.

Understand a screenshot

Ask for the visible steps, error wording or interface elements. Include the app name and version, and remove names, emails or account details first.

Work through a report

Ask for a section-by-section summary, key decisions, action owners or a list of statements that need checking against the original page.

Compare two documents

Define the comparison criteria and ask ChatGPT to label each difference with its document, heading and page or section.

Explore a spreadsheet

Ask for missing values, duplicates, trends, unusual results or totals grouped by a named column. State the units and date range.

Create a chart

Name the columns, grouping and chart purpose. Ask why that chart type is suitable, then inspect labels, scales and underlying values.

A Reliable Method

Inspect → Ask → Check → Refine

Good file work is a conversation, not a one-click answer.

Inspect: remove sensitive material, confirm permission and choose the clearest file format.
Prepare: crop an image, export a Google document to a supported format or tidy spreadsheet headers and rows.
Attach: use the attachment control in the message area and wait for the file to finish loading.
Ask: state the task, context, exact scope, output format and what uncertainty should be flagged.
Check: compare names, quotes, rows, totals, dates and conclusions with the original material.
Refine: correct assumptions and request a narrower section, clearer method or different output.

1 · InspectIs the file safe, clear and appropriate to upload?

2 · AskWhat exact result do you need, and for whom?

3 · CheckWhich source passages or calculations support the answer?

4 · RefineWhat needs correcting, narrowing or explaining?

Prepare for Better Results

Make the information easy to read

Input	Prepare it	Prompt for evidence	Verify yourself
Image	Use a sharp, upright image; enlarge small text; crop irrelevant areas; mark the target region.	“Describe only what is visible and label uncertain details.”	Text, counts, colour-coded lines, spatial positions and fine detail.
Document	Prefer selectable digital text; use clear headings; split very large or complex files into focused parts.	“Give the section heading and page for each finding.”	Quotes, page references, tables, footnotes and visuals embedded in the file.
Data	Use descriptive headers, one record per row and one main table; remove blank separator rows and explain units.	“Show the method, assumptions and rows excluded from the calculation.”	Row counts, missing values, filters, formulas, units, totals, chart axes and outliers.

Exact values need structured files

ChatGPT may not reliably extract exact values from image-based tables, scanned pages or complex visual layouts. When exact numbers matter, upload the original spreadsheet or a clean text-based file instead of a screenshot.

Copy-Ready Prompts

Start with a precise request

Replace the words in square brackets, attach the relevant file and then check the result.

Neutral image descriptionSeparate observation from inference.

Practical visual feedbackGet focused, actionable suggestions.

Evidence-linked summaryKeep the answer grounded in the file.

Document comparisonTrace differences to their locations.

Data quality inspectionCheck structure before calculating.

Transparent data analysisRequest the method as well as the answer.

Upload Readiness Check

Check before you attach

Tick only what is true for the file you plan to use.

PermissionI own this material or have permission and a lawful reason to use it.PrivacyI removed personal, confidential, financial, health and account information that is not essential.ReadabilityThe image is clear, the document has selectable text or the data table is tidy.Clear taskI can state exactly what I want ChatGPT to do and what it must not assume.OutputI know the audience, format, length and level of detail I need.VerificationI can compare the result with the source file, calculations or another trusted method.

Do not upload first and think about privacy later

When the material is sensitive or regulated, follow your organisation’s approved tools, policies and retention requirements. Use anonymised or fictional practice material for this course activity.

Common Limitations

What can go wrong

Visual misreading

Small, rotated, blurred or non-Latin text can be read incorrectly. Colour-coded graphs, object counts and precise positions may also be unreliable.

Missing embedded visuals

Some plans and document types use text-only retrieval, so images inside documents may be discarded. Upload important visuals separately where permitted.

Incomplete long-file review

A large or complex file may upload successfully without every part being analysed. Ask about named sheets, rows, columns or sections.

Incorrect method

ChatGPT may choose a calculation or chart that does not fit your purpose. Specify the method or ask to review and adjust it.

Confident fabrication

A polished answer can contain invented quotes, page references, totals or interpretations. Open the original and verify important details.

High-stakes misuse

Do not rely on image or file analysis alone for medical diagnosis, legal decisions, safety-critical work or significant financial choices.

Practical Activity

Complete a safe file challenge

Use a non-sensitive image, short document or small practice spreadsheet. Your progress is stored in this browser.

1. Choose a safe practice fileI selected material I am permitted to use and removed unnecessary personal information.2. Identify its input typeI decided whether this task needs image interpretation, document extraction or data analysis.3. Prepare the fileI improved readability, structure or focus before attaching it.4. Write a precise requestI included the task, scope, context, output format and boundaries.5. Attach and reviewI confirmed the correct file loaded before sending my request.6. Ask for evidenceI requested visible details, document locations or calculation methods.7. Verify the resultI checked at least three details against the original material.8. Refine onceI corrected an assumption or requested a clearer, narrower or more useful output.

Finish the Lesson

Record your progress

Mark Lesson 14 complete

Your choice is saved in this browser. The next lesson explores voice, Canvas and other ChatGPT tools.

Official References