Record the Origins of AI-Drafted Tables and Example Data

Document a table at the level needed to distinguish its generated presentation from the origin of its facts and example values. Record which entries are sourced, which are illustrative, and which wording was generated. A clean table layout does not prove the data is real, accurate, or supported by the cited source.

Identify what the table is meant to represent

Start by stating whether the table reports factual information, compares editorial observations, or demonstrates a fictional example. These purposes need different evidence and labeling. A table can look authoritative simply because its rows and columns are orderly, so the author should establish its role before reviewing individual values.

For example, a planning guide might include an invented schedule to show how tasks can be organized. That is different from a table claiming to report typical completion times collected from real projects. Both may contain plausible numbers. The manuscript should make the distinction clear, and the production record should explain where those values came from.

Separate structure, wording, and underlying information

Record whether AI helped propose column headings, draft explanations, generate example entries, or transform supplied data into a table. These are different activities. A model can organize verified information while also adding unsupported details, so a general note saying the table used AI is not enough for an accuracy review.

Keep the underlying source material available separately from the rendered table. Identify which source supports each factual group of entries where necessary. This lets the author check the transformation rather than assuming that a citation at the bottom validates every cell. Preserve the distinction between an accepted source and the generated interpretation of that source.

Symbolic publishing workflow for separate structure, wording, and underlying information
Separate structure, wording, and underlying information

Label illustrative values so they are not mistaken for findings

When the table uses invented demonstration data, make that role clear in the manuscript. A reader should not have to infer it from the absence of a citation. Use the minimum detail needed to teach the task and avoid presenting a fictional sample as though it were a measured average, surveyed result, or typical outcome.

Check the surrounding prose for language that changes the table's status. A caption may correctly call it an example while the next paragraph refers to what the data proves. Resolve that contradiction. The example can illustrate a method without providing evidence about how frequently an outcome occurs outside the imagined scenario.

Verify factual entries against the accepted sources

Compare names, categories, values, and qualifications with the material they are supposed to represent. Look for omitted conditions, changed units, or added conclusions. A generated table may simplify a source in ways that make its rows easier to compare while removing distinctions that matter. Review those editorial choices explicitly.

Do not fill a missing entry with a plausible value merely to create a complete-looking table. Mark the information as unavailable or revise the table's design when appropriate. The author should prefer an understandable limitation to invented precision. Keep unresolved cells visible in the working record until they have a supported disposition.

Symbolic review checklist for verify factual entries against the accepted sources
Verify factual entries against the accepted sources

Preserve creation history through revision

KDP distinguishes AI-generated content from assistance applied to content the author created, and its guidance retains the generated distinction after substantial editing of generated material. Keep the actual process recorded for retained table text and other generated content. An accuracy check or a manual formatting pass does not replace that history.

When a table is rebuilt, note which material remains from the earlier version and which was replaced. Retain the source references and example-data designation through that change. A new file or a different visual style should not make a fictional example appear to be newly collected evidence or separate factual entries from their original support.

Review the final table with its explanation

Inspect the exported table, caption, notes, and surrounding discussion together. Confirm that units and qualifications remain visible and that the table's purpose is still clear after layout. A source note lost across a page break can weaken the reader's understanding even when the main values remain intact.

Save the accepted table source, evidence register, origin record, and final rendering with the release package. These records support different questions: what the table means, whether its claims are supported, and how its retained content was created. Keeping them distinct makes later corrections more reliable and prevents a polished presentation from standing in for evidence the project never collected.

Frequently asked questions

Clear answers for this publishing decision.

Does citing one source under a table validate every entry?

No. Check which entries the source actually supports and whether generated transformations added or changed information. Keep unsupported cells visible until resolved.

Can invented values be used in a teaching example?

Clearly identify them as illustrative and avoid using them as evidence for real-world frequency or outcomes. Review captions and surrounding claims for consistency with that status.

Does manually checking a generated table remove the need for an origin record?

No. Accuracy review and creation history answer different questions. Preserve both for the retained publication material and apply current KDP guidance.

AI use and quality control

What must I check, disclose, and verify myself when AI helped produce the book?

Review facts, rights, and source material yourself, record which parts were AI-assisted or AI-generated so you can answer Amazon’s disclosure question accurately, and treat every automated check as a starting point rather than a verdict.

See what the compliance checks cover

Get all of our updates directly to your inbox.
Sign up for our newsletter.

We could not save your subscription.