What it means
OCR is technology that converts images of printed or typed text, such as scans and photos, into machine-readable text. It is often the first step in making paper documents searchable and usable by AI systems.
Real-world example
A district court digitises 40 years of printed annual reports. OCR turns the scanned pages into text, but tables, faded copies and old typefaces produce errors, so the team measures error rates on samples and corrects the most important documents before adding them to the search system.
Why it matters for AI adoption
Many organisations' knowledge is locked in scanned documents. OCR quality directly affects what downstream AI systems can find and how accurately they answer. Errors in numbers, names and tables are especially risky.
Often confused with
Handwritten Text Recognition (HTR)
OCR is designed for printed or typed text. Handwriting requires Handwritten Text Recognition (HTR), which uses models trained on handwriting styles and usually needs more correction.
Questions to ask
- How accurate is OCR on our actual documents, including tables and older material?
- How will OCR errors be detected and corrected for important documents?