OCR makes sense. I use it a lot to find indexed news articles from newspaper archives, and the “spelling” mistakes are often very bizarre.
OCR doesn’t work that well sometimes. Sometimes it is faster for a human being to transcribe the document.