Hell USPS has been using machine learning (yes a kind of AI but not the kind they are implying) for years to do that kind of thing.
Comment on Block ditches 4,000 staff, because AI can do their jobs
CanadaPlus@lemmy.sdf.org 3 weeks agoEh, I know this is the anti-AI instance, but reading and interpreting things like that is something you can verifiably get AI to do 90% of the time.
scintilla@crust.piefed.social 3 weeks ago
Paradox@lemdro.id 3 weeks ago
Kind of
They’ve had several address resolution centers around the country, where reviewers look at mail and figure out it’s address. They don’t physically handle the mail, it’s an image on a screen.
Iirc they’ve been doing it this way since the 70s
scintilla@crust.piefed.social 3 weeks ago
No? For everything they can they just use OCR and then send it on its way without a human having ever seen it sometimes. If the hand writing is bad enough that the machine can’t figure it out that’s where the human reviewers come in.
Ludicrous0251@piefed.zip 2 weeks ago
Until you realize those 10% errors missed the decimal point.
CanadaPlus@lemmy.sdf.org 2 weeks ago
Again, that’s what the 6000 remaining employees would be for.
Powderhorn@beehaw.org 3 weeks ago
Because 90% accuracy is acceptable for financial institutions …
XLE@piefed.social 3 weeks ago
I’ve got an idea. If 90% of AI’s output is accurate, just have humans review the 10% that will be inaccurate.
(Yes I am an AI expert, how did you know)
TehPers@beehaw.org 2 weeks ago
Which outputs are accurate, and which ones are inaccurate? How could you tell? What steps did you take to verify accuracy? Was verifying it a manual process?
XLE@piefed.social 2 weeks ago
That’s easy. You just get a second AI to ask the first AI if their responses were accurate or not
(/s)
CanadaPlus@lemmy.sdf.org 3 weeks ago
No, it’s really not. Thus the 6000 remaining employees.
(Assuming this is a significant part of their business)