Part 1 · Know Your AI Readiness
Data · What Good Data Looks Like
You have handed generative AI the repetitive parts of your week. Now you feed it last quarter's sales file, three date formats, blank cells, and "N/A" typed five different ways, and it returns a clean, confident, wrong answer.
What good data looks like
Artificial intelligence does not know when your data is wrong. It does not pause to say "these dates look inconsistent." It faithfully finds patterns in whatever you give it, and when the input is noise, the output is confident noise. So the skill here is not to trust AI less. It is to look at your own data the way the machine will.
Your data arrives in two shapes, and "good" means something different for each. Structured data is rows and columns: a spreadsheet, an export, a table. Good structured data is consistent (one date format, one unit, one spelling for each category), reasonably complete, accurate against reality, and free of duplicates. A column that mixes "Singapore", "SG", and "S'pore" reads as three different customers to a machine.
Documents and text are everything else: reports, emails, PDFs, meeting notes. For these, good means the AI can read the file itself (selectable text, not a photograph of a page), you have the current version rather than last year's draft, the content stays on topic, and the source is one you trust.
| Structured data | Documents & text |
|---|---|
| Consistent One format per column | Readable Real text, not a scan image |
| Complete Few blank cells | Current The right version |
| Accurate Matches reality | Relevant On topic |
| Unique No duplicates | Trusted A source you rely on |
Key Takeaway
AI faithfully finds patterns in whatever you give it, so the quality of what you give it is on you.
Check your understanding
Self-check only. Nothing is scored and nothing is recorded; open an answer to compare it with your own.
A colleague asks AI to summarise "our refund policy" and uploads a document, but it is last year's version. The summary looks clean and confident. What does this illustrate about data quality?
- Nothing, if the summary reads well, the data was fine.
- Document quality matters as much as spreadsheet quality: feeding AI an out-of-date source produces a confident answer for the wrong reality.
- This is fine because text data does not need to be accurate, only spreadsheets do.
- The AI made a mistake; the document was irrelevant to the result.
Show the answer
Document quality matters as much as spreadsheet quality: feeding AI an out-of-date source produces a confident answer for the wrong reality.
Which statement best reflects the AI-Ready (L2) understanding of data quality?
- "As long as my spreadsheet is clean, the AI's answer will be reliable."
- "Data quality is the IT team's problem, not mine."
- "Both my structured data and my documents need to be complete, current, and accurate, because the AI's answer is only as good as what I give it."
- "Modern AI is smart enough to fix bad data on its own."
Show the answer
"Both my structured data and my documents need to be complete, current, and accurate, because the AI's answer is only as good as what I give it."
Saved in this browser only. Carry progress between devices with the resume link on the course page.