Documents, spreadsheets, and multimodal work
Check AI-created documents and charts against the original data.
In this lesson · A5 of 8
All lessons in this track →Progress for this track is saved in this browser only. No completion badge is issued. View your learning.
Start with the idea
Check AI-created documents and charts against the original data.
AI can misread an image or copy a number incorrectly. Compare the extracted data with the source before calculating. Keep the original and record corrections.
OCR: Software that turns text in an image into readable text. It can make mistakes.
See the example and check your understanding
A made-up teaching example
The source lists 80, 20 and zero confirmed attendees. The draft reads the first number as 30. The correct total is 100, and the first group is 80 percent of it.
Ask yourself
Does zero confirmed attendees mean the venue has zero capacity?
Show the explanation
No. Attendance and capacity are different facts. Correct both the copied number and calculations that depended on it.
Start with these checks
- Check copied values against the source.
- Recalculate totals and percentages.
- Explain what zero and unknown mean.
Use the different practice case below and check every requirement in its full instructions.
The video explains the main idea. Use the sources, instructions and checks below to complete the practice.
Captions appear here during playback.
Documents, spreadsheets, and multimodal work
Check AI-created documents and charts against the original data.
Video playback is unavailable. Read the descriptive transcript below, go directly to practice, or open the video file.
Source-based teaching example · Fictional data · Reviewed 2026-10-03. Microsoft Mark · locally synthesized. On a small screen, use fullscreen for diagram details or read the transcript.
Read the descriptive transcript
0:00 · What you will learn
Check AI-created documents and charts against the original data.
What appears on screen
Check AI-created documents and charts against the original data.
0:06 · The idea
AI can misread an image or copy a number incorrectly. Compare the extracted data with the source before calculating. Keep the original and record corrections.
What appears on screen
AI can misread an image or copy a number incorrectly. Compare the extracted data with the source before calculating. Keep the original and record corrections.
0:20 · One useful term
OCR. Software that turns text in an image into readable text. It can make mistakes.
What appears on screen
Software that turns text in an image into readable text. It can make mistakes.
0:29 · A worked example
The source lists 80, 20 and zero confirmed attendees. The draft reads the first number as 30. The correct total is 100, and the first group is 80 percent of it.
What appears on screen
The source lists 80, 20 and zero confirmed attendees. The draft reads the first number as 30. The correct total is 100, and the first group is 80 percent of it.
0:44 · Your turn to think
Does zero confirmed attendees mean the venue has zero capacity? Pause and decide before continuing.
What appears on screen
Does zero confirmed attendees mean the venue has zero capacity? Pause the video if you need more time.
1:01 · Why this answer works
No. Attendance and capacity are different facts. Correct both the copied number and calculations that depended on it.
What appears on screen
No. Attendance and capacity are different facts. Correct both the copied number and calculations that depended on it.
1:13 · Check 1
Check copied values against the source.
What appears on screen
Check copied values against the source.
1:18 · Check 2
Recalculate totals and percentages.
What appears on screen
Recalculate totals and percentages.
1:24 · Check 3
Explain what zero and unknown mean.
What appears on screen
Explain what zero and unknown mean.
1:29 · Now try a different case
Open the lesson and try its different practice case. Use every check in the full task. Watching alone does not mark the lesson complete.
What appears on screen
Open the lesson Try the new case Check and revise The full task and its requirements are in the linked lesson.
Watching explains the example. Complete the lesson practice to check what you can do. Explore all visual lessons →
Before you start
Prerequisite: A3–A4 source provenance and grounding. Offline/no-cost route: paper, calculator and the synthetic facts below. Optional tools must already be authorized. No paid API, private workplace record, account creation or external action is required.
What you will be able to do
- Transform without inventing values.
- Recompute calculations and verify OCR/image claims.
- Preserve accessible structure and change history.
Part 1 Use a check matched to the medium
For prose, compare meaning and authority against the source. For spreadsheets, recompute from original values and inspect formulas, units, denominators, blanks and rounding. OCR extraction is a hypothesis about visible text; compare critical fields with the image. Multimodal model support varies: image input does not imply image generation or dependable small-text reading. Distinguish observed content from interpretation and uncertainty. Accessible outputs require useful headings, logical reading order, labels, equivalent text/table routes and meaningful descriptions; visual polish alone establishes neither accuracy nor accessibility. Preserve source and corrected versions separately.
Part 2 Worked synthetic example
Synthetic workshop table: A=80 attendees, B=20, C=0 confirmed. OCR draft reads A=30 and total=50. Original source transcription supplied for offline work says 80/20/0. Recompute 80+20+0=100; A share is 80/100=80%. Zero confirmed is not zero capacity. Correct both the OCR value and total, preserving the original error record. A chart description should say A=80, B=20, C=0 confirmed, not all venues are empty or available.
Part 3 Guided practice and feedback
Match methods: memo obligation -> source sentence comparison; total -> source-cell calculation; scanned value -> image comparison (or supplied authoritative transcription offline); chart trend -> underlying data and axes; accessibility -> structure/reading route review. A model-generated explanation of its own total is not independent recomputation.
Review and next step
Self-review is learning evidence, not independent certification or operational approval. A6 evaluates whether these checks improve the complete workflow.
Substantive review: 2026-10-02. See the configuration guide and the NIST AI RMF, a voluntary framework rather than a course authorization or credential.