Let the robot do the drudgery
At the moment, I’m trying to get a clean OCR of a long and often dry early Victorian history book. Luckily for me, someone else did the scans – but the original OCR was awful. I’m doing it page at a time and it’s scooting along pretty well – except for the dreadful tables.
That’s them in the thumbnails above.
Then I had an idea – I wonder if ChatGPT will do those? Spoiler: yes. I upload the scan, tell it how I want the table formatted (fonts, etc.) and it gives me a link to a docx file I can just drop in.
At one point, I got too clever for my own good – I was giving it the big, raw png files – and I hit my data limit. I’ve never hit my limit as a paying customer! But it reset within a few minutes and I was told that wouldn’t happen if I jpg’ed them all. It even told me how to batch jpg the whole directory.
Some would say beware – before long, it will do the whole job for me. But y’know, I would be totally okay with that.
March 31, 2026 — 5:51 pm
Comments: 4