is Artificial intelligence (AI) revolutionizing gemology? or potential job displacement? labs emphasize that AI complements human expertise rather than replacing it.
thoughts?
Sarine: Digital & Virtual Diamond Grading Now Just a Step Away
GIA and IBM Research Join Forces to Transform Diamond Grading
Gübelin artificial intelligence programme to provide greater consistency in gemmology grading
AI in Gemology! Innovations or Concerns?
Moderators: Stephen Challener, Barbra Voltaire, FGG, Alberto
- roshanravan
- Gemology Online Veteran
- Posts: 950
- Joined: Thu Feb 22, 2007 10:23 am
- Location: NYC
- Contact:
AI in Gemology! Innovations or Concerns?
Farshid Roshanravan
- S. Rautiainen
- Active Member
- Posts: 93
- Joined: Tue Dec 01, 2020 10:16 pm
- Location: Canada
Re: AI in Gemology! Innovations or Concerns?
I tried out ChatGPT 3.5 with some gemmology experiments. In one series, I had it write up some gemmology articles and they weren't bad though a little sparse on the finer details. In general, if you're needing an article on known information it wasn't bad, but you'd have to adjust it.
A different series was where I had it be my "lab assistant" and help identify the stones. I would tell it the basic observations and ask what tests would help identify the stone. It provided the tests it asked for and then I would provide the results of those tests. It would then identify the stone or ask for more results. Besides one time where it couldn't properly identify Peridot, it was actually pretty accurate. It was even able to identify some pretty obscure stones I had laying around so I personally think, while it won't replace humans doing the tests completely (at least not yet) the AI would be certainly helpful to the average gemmologist if it was trained properly.
I've attached a snip of one of my later tests (where I just provided some test results and asked it to help identify the stone) in case you're curious to see what we have so far.
A different series was where I had it be my "lab assistant" and help identify the stones. I would tell it the basic observations and ask what tests would help identify the stone. It provided the tests it asked for and then I would provide the results of those tests. It would then identify the stone or ask for more results. Besides one time where it couldn't properly identify Peridot, it was actually pretty accurate. It was even able to identify some pretty obscure stones I had laying around so I personally think, while it won't replace humans doing the tests completely (at least not yet) the AI would be certainly helpful to the average gemmologist if it was trained properly.
I've attached a snip of one of my later tests (where I just provided some test results and asked it to help identify the stone) in case you're curious to see what we have so far.
FCGmA, Owner.