Wow, I'm not surprised as imaging can look at a coin dozens or hundreds of times per second and analyze every nick and bag mark and measure their depth to fractions of a millimeter or microns. No way the best grader can do that giving 15-20 seconds per coin. The human eye can only see so much in 1-2 seconds. Maybe they spend more time on more expensive or special coins (i.e., that 1794 Silver Dollar or an MCMVII UHR Saint-Gaudens) but that really just proves my point. Don't get me wrong...I'm glad we have the TPGs and I think overall they do a good job and have been a net-positive for the hobby. But I am sure AI along with 4K or 8K imaging systems will be able to do the equivalent of an MRI on a coin in seconds. Might even be able to tabulate and graph all the bag marks and other blemishes with size...length...width...and even DEPTH !!
I won't claim to be an expert on AI but from what I gather, we're rapidly approaching a future where robots are replacing manual labor and AI is gradually replacing human creativity. I'm not saying this is great or I'm really happy about it. Its just happening and can't be ignored. People in this thread have greatly under estimated how rapidly this stuff is evolving. Yes, some of the basic programs online are simply utilizing existing data that's already out there and it's not exactly full blown AI. It's not that impressive. Yet. But to think a computer along with AI won't be able to grade better than humans in the very near future isnt thinking realistically. There doesn't need to be emotion to grade accurately. In fact you probably want to leave the emotion out of it. (I'd rather not have somebody grading my stuff on a Monday after a rough weekend.) Computer scanners will be able to do this and they'll be able to grade better and more consistant, than humans can. The computing power that is rolling out now is difficult to wrap our heads around. Not that long ago, a basic flash drive was only 256 MB. I remember when 1 gig came out and that was a lot of storage. They were expensive. Then 256 Gig, 512 Gig. Now you can get a 2 Terabyte flash drive (2k gig) of data on a simple flash drive for $20 from Walmart. Storage capacity like that used to cost hundreds of dollars just a few years ago. The capacity a grading computer would be able to have is infinite and instantly accessible. It would never need breaks or days off. It will have the capacity to store every coin of every type that ever goes through it in infinite detail. It will likely be able to compare say a 1964 quarter against all other 1964 quarters it has seen, plus every grading guideline in existence, in mere seconds if not fractions of a second. It'll be connected to a mainframe that has data that other computers have stored. If anything the grading will likely get far more accurate and more detailed while also getting a lot more grading done. Counterfeits and cleaning will be more accurately detected. Hell, they'll proabably have real time metalergic analysis while grading is underway. In 2018 we had a super computer that was considered the best of the best at the time. In 2024 they had a new one at a lab in CA that they say is 22x better than that machine from 2018. Uses 30 megawatts of electricity and runs 1.742 quintillion calculations per second! Complex high res 3D simulations that would have taken weeks or months on the 2018 platform are done in hours or days on this new one. And on it goes. This just keeps unfolding and getting better. I'm not saying NGC and PCGS are going to have a supercomputer any time soon but this massive capability and advancement eventually trickles down. There will be a desire to take advantage of this computing power. Grading seems like the ideal example of a task that will totally be taken over.
That may be factual, and a fair analysis, but (there is always a "but"), there is AI residing in my home that still does not remember what day the damn trash is picked up, no matter how many times I programmed the data input.