Wrong dates and math can wreck a project

A number with decimals or a full calendar date signals care. AI can still:

• shift a famous date by a day or a year
• convert units incorrectly (km vs miles)
• misread percentages (“40% of 30”)
• add steps that cancel an earlier error but still look neat

Treat every critical number as guilty until recalculated or re-looked-up.

Prioritize checks for numbers that:

• appear in your title or thesis
• affect a grade rubric (“include correct historical date”)
• change a chart or conclusion
• could mislead your audience (health, money, safety — even in school projects)

Decorative stats you barely use can wait. Load-bearing numbers get a human recompute.

For math:

1. Cover AI’s final number.
2. Recompute with paper, calculator, or spreadsheet.
3. Compare.
4. If mismatched, find the first bad step — don’t just patch the ending.

For dates:

1. Check class notes or a trusted timeline.
2. Confirm time zone / calendar quirks only if your assignment cares.
3. Match spelling of names tied to the date.

Show each calculation step. Restate units. Do not skip arithmetic.

Helpful for spotting breaks — but you still redo the key line. Models can narrate wrong math in complete sentences.

One bad percentage becomes a huge wrong pie slice. One off-by-one date becomes a timeline arrow pointing at the wrong decade. Before you lock design, print or screenshot the slide and mark every numeral with a pen. If you cannot point to where that numeral came from — your survey raw count, your textbook, your recalculation — it does not deserve pixels. Pretty layout cannot rescue a false digit.