You could’ve just looked for off the shelf OCR software and it would probably be better, no LLM needed. OCR has been around for far longer than the current LLM bubble.
Comment on Data Shows That AI Use Is Now Declining at Large Companies
jaykrown@lemmy.world 1 day ago
Skill issue. I used AI to create a web application that extracts the serial number from an image into text. This allows us to just simply take a picture rather than than having to type the serial number manually while using a magnifying glass. Significantly speeding up the process and lowering error rate.
AmbiguousProps@lemmy.today 23 hours ago
jaykrown@lemmy.world 22 hours ago
No, I tried OCR and it was less accurate.
ganryuu@lemmy.ca 4 hours ago
You’re reading text from a picture. That is OCR.
Jhex@lemmy.world 1 day ago
indeed, anyone with skills would have whipped that up in noetime without AI… or use any of the many apps that already does that
iamericandre@lemmy.world 1 day ago
For real, you can copy text out of images on most modern phones so this isn’t really an issue anymore
jaykrown@lemmy.world 22 hours ago
Yea, you can do that, but have fun doing that 1000 times a day. Manually copying was the problem, I created a system that automated that process.
DivineDev@piefed.social 1 day ago
Ok but how does the phone extract the text? The fact that a modern phone can do it does not mean there’s no AI involved
Blaster_M@lemmy.world 1 day ago
OCR is a very, very old AI system that for decades we’ve scanned documents and saved to pdf with. The old system is far from perfect, but it does work well enough in most cases. If an AI recognition model can do the same job with less mistakes, that is an improvement.
jaykrown@lemmy.world 22 hours ago
Then why was I the one to do it if anyone could have done it?
TheBat@lemmy.world 3 hours ago
en.wikipedia.org/wiki/Dunning-Kruger_effect