Well, maybe.
Let me give you a brief tour of just a few of the challenges you'll be up against.
Size - Until die cutting became the standard, there is some size variance on all cards. So while you can measure size accurately with existing technology, you still have to determine what the allowable tolerance is. Too much and people will get the impression you miss trimming. Too little and you'll turn customers off and get very little business.
Edge quality - A major challenge. And the biggest factor in spotting trimming. So for each set, the computer will have to know how it was cut. And while many modern cards are die cut, some have edges that appear crisp(1988 score), while others look more like a T206 with the beveling and ridge. (Most recent Gypsy Queen) Some cards in some sets have BOTH sorts of cuts. (Some but not all cards in the Gypsy queen set from I think 2014 or 2015 maybe others I'm too lazy to dig the box out. )
How do you treat cards that were perforated? Like 51 Topps. Are the little torn nibs counted as damage or a proper edge?
And even if the card looks good, can you scan it quickly at high resolution to see the tool marks that should be on each edge.
I have a hard time spotting them under 40X magnification, and they're not always obvious.
Surface - Under high magnification, cardstock isn't actually "smooth" Telling a spot where a fiber came out from a scratch isn't always easy. And seeing some stuff like that depends on the angle of the light. I don't know of an existing 3d scan technology with enough resolution to even detect a slight scratch.
And older cardboard sometimes has what are called "inclusions" or bits of stuff that got into the cardstock during manufacture. One of my T206s has a small but obvious inclusion, which I believe kept it from getting a high grade. The white spot is the area of the inclusion, and the raised are caused ink loss, but not paper loss. After some thought I figured the 40 was a fair grade. But that's a judgement call, as the inclusion has been there since before it was a card.
Modern glossy cards have even more complications. Some gloss comes out very smooth, some doesn't. That slight difference is visible on cards from at least the 70's until now. (Some Topps from the 70's into early 80's have a gloss that yellows over time, some a re glossier, but I can't tell if it's smoother because there's more, or if the cardstock absorbed some or if the actual gloss is different. Some very glossy cards can be found with gloss that isn't smooth, but should be. Probably a result of the gloss being sticky in the press and leaving tiny raised points. (Sort of like textured paint) Telling that apart from damage I believe would be a challenge for a machine, not so much for a person.
Or.....93 upper deck, where the gloss was applied three different ways on about 1/3 of the set. Picture only, entire back, and picture only with the entire back done later. It's visible, but by a machine? Maybe maybe not.
And will the machine kick back all the batter ups because of the die cutting in the center of the card?
So you not only need some scanning technology that doesn't currently exist, you need a database of the minutia of every card set you'll try to do. And software that can compare what's being scanned to the database. Not "hard" but that will be a HUGE database, and people will have to enter the information. And unless you program it to identify every card in every set, even through what might be very heavy damage, people will have to tell the machine what it's looking at.
Oh yeah, and you'd need to constantly upgrade it as technology improved and operating systems changed.
And that's only the tip of the iceberg I believe the SS computergraded Titanic will run into.
Or someone could train a kid who likes old stuff for maybe a week, have them start with cheap commons to get a feel for things, and gradually teach them more. Whoever it was would be much faster, but wouldn't be 100% accurate. They'd probably cost a lot less too.
Steve B