Home U.S. Coin Forum

AI just broke through a major bottleneck in coin analysis

13

Comments

  • DocBenjaminDocBenjamin Posts: 1,491 ✭✭✭✭✭
    edited October 23, 2024 11:36AM

    First Ebay buy was a rare record from a guy in Japan. Late 2000.

    Thought it was cool to get an email from the other side of the globe in a split second. Problem is, now I get a dozen a day.

    :/

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @DocBenjamin said:
    First Ebay buy was a rare record from a guy in Japan. Late 2000.

    Thought it was cool to get an email from the other side of the globe in a split second. Problem is, now I get a dozen a day.

    :/

    Only a dozen?

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @Typekat said:
    I’m just trying to “remember” the Internet in 1980…

    It was restricted largely to national labs and the like. It was all text at a couple hundred baud on dial up modems.

    BS! Everyone know that Al Gore invented the Internet.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭
    edited October 24, 2024 5:07AM

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

    That is not what I said. What i said is that the AI COULD see a billion data points on EACH coin.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭
    edited October 24, 2024 4:44AM

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

    That is not what I said. What i said is that the AI COULD see a billion days points on EACH coin.

    Sorry. I think that would be called "micro-grading" in a class. It is discouraged. Perhaps AI should be set at 10 days points to be like a 5X magnifier and closer to a professional grader and market acceptability.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

    That is not what I said. What i said is that the AI COULD see a billion days points on EACH coin.

    Sorry. I think that would be called "micro-grading" in a class. It is discouraged. Perhaps AI should be set at 10 days points to be like a 5X magnifier and closer to a professional grader and market acceptability.

    The developer can do whatever they want. That is the point. They can grade without looking at the coin at all, if you want. You use whatever algorithm gives you the desired precision.

  • ConnecticoinConnecticoin Posts: 12,895 ✭✭✭✭✭

    My sense is this will fade away along with the promises of “computer” grading in the early 00s. Among other things, I highly doubt AI can factor in eye appeal the way it is done with the current TPGs. AI cannot even give a decent coin description - the AI coin descriptions on ebay stand out like a sore thumb - like a badly plagiarized term paper.

  • TypekatTypekat Posts: 416 ✭✭✭✭

    That’s roughly the timeline that I thought.

    30+ years coin shop experience (ret.) Coins, bullion, currency, scrap & interesting folks. Loved every minute!

  • logger7logger7 Posts: 8,567 ✭✭✭✭✭

    @jmlanzaf said:

    @Typekat said:
    I’m just trying to “remember” the Internet in 1980…

    It was restricted largely to national labs and the like. It was all text at a couple hundred baud on dial up modems.

    A retired programmer I know is highly skeptical of the cultural advantages of the internet despite the fact that he virtually lives on it; just way too much cultural fragmentation with all the information. My understanding is that it came out of the military that needed a dedicated form of communication that it provides.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @logger7 said:

    @jmlanzaf said:

    @Typekat said:
    I’m just trying to “remember” the Internet in 1980…

    It was restricted largely to national labs and the like. It was all text at a couple hundred baud on dial up modems.

    A retired programmer I know is highly skeptical of the cultural advantages of the internet despite the fact that he virtually lives on it; just way too much cultural fragmentation with all the information. My understanding is that it came out of the military that needed a dedicated form of communication that it provides.

    That was the first application although, if i recall, it was originally developed as an internal network at Xerox PARC.

  • messydeskmessydesk Posts: 20,007 ✭✭✭✭✭

    @jmlanzaf said:

    @logger7 said:

    @jmlanzaf said:

    @Typekat said:
    I’m just trying to “remember” the Internet in 1980…

    It was restricted largely to national labs and the like. It was all text at a couple hundred baud on dial up modems.

    A retired programmer I know is highly skeptical of the cultural advantages of the internet despite the fact that he virtually lives on it; just way too much cultural fragmentation with all the information. My understanding is that it came out of the military that needed a dedicated form of communication that it provides.

    That was the first application although, if i recall, it was originally developed as an internal network at Xerox PARC.

    Ethernet, yes. The internet, no.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭
    edited October 24, 2024 10:14AM

    @messydesk said:

    @jmlanzaf said:

    @logger7 said:

    @jmlanzaf said:

    @Typekat said:
    I’m just trying to “remember” the Internet in 1980…

    It was restricted largely to national labs and the like. It was all text at a couple hundred baud on dial up modems.

    A retired programmer I know is highly skeptical of the cultural advantages of the internet despite the fact that he virtually lives on it; just way too much cultural fragmentation with all the information. My understanding is that it came out of the military that needed a dedicated form of communication that it provides.

    That was the first application although, if i recall, it was originally developed as an internal network at Xerox PARC.

    Ethernet, yes. The internet, no.

    Yes. But it had is roots in the same data transfer technology, did it not?

  • mr1931Smr1931S Posts: 6,252 ✭✭✭✭✭

    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Great spirits have always encountered violent opposition from mediocre minds.-Albert Einstein

  • fathomfathom Posts: 1,723 ✭✭✭✭✭

    Essentially was said 4 decades ago about TPGs.

    The marketplace will decide if robograding is viable.

  • AUandAGAUandAG Posts: 24,782 ✭✭✭✭✭

    Although I invest in AI, I don't believe it should apply to coins. Which AI will be the "best" or "cheapest". Do you get what you pay for? Too many questions and then there's the addage: If it ain't broke, don't fix it.
    bob :)

    Registry: CC lowballs (boblindstrom), bobinvegas1989@yahoo.com
  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Substitute TPGS for robot and you can ask the same question.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @AUandAG said:
    Although I invest in AI, I don't believe it should apply to coins. Which AI will be the "best" or "cheapest". Do you get what you pay for? Too many questions and then there's the addage: If it ain't broke, don't fix it.
    bob :)

    It would be the NGC AI or the PCGS AI.

  • coastaljerseyguycoastaljerseyguy Posts: 1,379 ✭✭✭✭✭

    If NASA is testing the use of AI on some space flight modeling and already is using to identify mineral composition on Mars, & AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.
    If the best graders in numismatics identified a # of examples of what constitutes a P01 to MS70 grade, PL/DMPL definition & examples, AT vs NT, known counterfeit data, coin series mintage strike & luster data, etc. and feed into a program, and applied quality lighting & photography, AI could easily apply such software to quality & consistent grading in the future.

  • messydeskmessydesk Posts: 20,007 ✭✭✭✭✭

    @jmlanzaf said:

    @AUandAG said:
    Although I invest in AI, I don't believe it should apply to coins. Which AI will be the "best" or "cheapest". Do you get what you pay for? Too many questions and then there's the addage: If it ain't broke, don't fix it.
    bob :)

    It would be the NGC AI or the PCGS AI.

    But an AI grading engine would not be static. It would be retrained as part of ongoing maintenance to make it "better." That would normally mean producing results closer to the ground truth more consistently. The issue here is that the ground truth is defined by fickle grading services and is rather elusive.

  • fathomfathom Posts: 1,723 ✭✭✭✭✭

    @AUandAG said:
    Although I invest in AI, I don't believe it should apply to coins. Which AI will be the "best" or "cheapest". Do you get what you pay for? Too many questions and then there's the addage: If it ain't broke, don't fix it.
    bob :)

    There is always room for improvement.

    It won't solve subjective disputes about eye appeal etc. but it has the potential to speed services and maintain a consistency, albeit an imperfect one.

  • LakesammmanLakesammman Posts: 17,393 ✭✭✭✭✭

    No matter how AI grades, I still have to like the coin.

    If AI can fix voice non-recognition, I'll be mildly impressed.

    "My friends who see my collection sometimes ask what something costs. I tell them and they are in awe at my stupidity." (Baccaruda, 12/03).I find it hard to believe that he (Trump) rushed to some hotel to meet girls of loose morals, although ours are undoubtedly the best in the world. (Putin 1/17) Gone but not forgotten. IGWT, Speedy, Bear, BigE, HokieFore, John Burns, Russ, TahoeDale, Dahlonega, Astrorat, Stewart Blay, Oldhoopster, Broadstruck, Ricko, Big Moose.
  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @Lakesammman said:
    No matter how AI grades, I still have to like the coin.

    If AI can fix voice non-recognition, I'll be mildly impressed.

    No matter how PCGS grades, i still have to like the coin.

  • mr1931Smr1931S Posts: 6,252 ✭✭✭✭✭

    @jmlanzaf said:

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Substitute TPGS for robot and you can ask the same question.

    Slabbing makes the product more marketable so that's in the best interests of the dealer trying to find a buyer.

    Great spirits have always encountered violent opposition from mediocre minds.-Albert Einstein

  • leothelyonleothelyon Posts: 8,464 ✭✭✭✭✭

    AI grading would go along ways for those who want their coins graded by an extremely strict grading standard. Or AI could be programmed to leave out certain aspects of the coin and give separate scenarios of grades. For example; AI is programmed to ignore the luster or the condition and cough up more grades.....oh wait.............

    Leo :D

    The more qualities observed in a coin, the more desirable that coin becomes!

    My Jefferson Nickel Collection

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @mr1931S said:

    @jmlanzaf said:

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Substitute TPGS for robot and you can ask the same question.

    Slabbing makes the product more marketable so that's in the best interests of the dealer trying to find a buyer.

    Non sequitur. Has nothing to do with what I said. Are you saying that you would buy a slabbed coin that you don't like?

    But, to your non sequitur, for all you know, robot slabbing might also create a more marketable product.

  • IkesTIkesT Posts: 3,199 ✭✭✭✭✭

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    That's harsche.

  • messydeskmessydesk Posts: 20,007 ✭✭✭✭✭

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

  • DocBenjaminDocBenjamin Posts: 1,491 ✭✭✭✭✭

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

  • DocBenjaminDocBenjamin Posts: 1,491 ✭✭✭✭✭

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

  • fathomfathom Posts: 1,723 ✭✭✭✭✭

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    Yup.

    At the end of the day it will take some significant time and resources to put together the mechanical, image and software plan to test. Then it will need to be refined. This is a monumental exercise for this type of marketplace. Not saying it isn't being developed as we speak but it is a huge investment.

    The chip systems and high performance computing hardware is already available to accomplish this task.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

  • AlbumNerdAlbumNerd Posts: 195 ✭✭✭

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

  • DocBenjaminDocBenjamin Posts: 1,491 ✭✭✭✭✭

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

    If the Google self driving software sees a bicycle and draws the incorrect conclusion, you have a corpse on a Tempe Arizona street.

    Try again.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

    If the Google self driving software sees a bicycle and draws the incorrect conclusion, you have a corpse on a Tempe Arizona street.

    Try again.

    So you're not intelligent because you never draw incorrect conclusions?

    Your confusing intelligence with omniscience.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

    That is not what I said. What i said is that the AI COULD see a billion days points on EACH coin.

    Sorry. I think that would be called "micro-grading" in a class. It is discouraged. Perhaps AI should be set at 10 days points to be like a 5X magnifier and closer to a professional grader and market acceptability.

    The developer can do whatever they want. That is the point. They can grade without looking at the coin at all, if you want. You use whatever algorithm gives you the desired precision.

    Sort of like picking a standard and nobody likes it or uses it. I wonder who the developer is going to be. Can't we all just get along...LOL! NUTS, to AI grading. It WILL happen but will not replace humans.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Because the profit makers in the grading business will continue to fleece the ignorant. Any fool can sell slabs and your coins better be slabbed to sell them easily. Ever here someone say: "Why isn't it graded? There must be something wrong with it?" In the future you an insert AI into the statement.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @coastaljerseyguy said:
    If NASA is testing the use of AI on some space flight modeling and already is using to identify mineral composition on Mars, & AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.
    If the best graders in numismatics identified a # of examples of what constitutes a P01 to MS70 grade, PL/DMPL definition & examples, AT vs NT, known counterfeit data, coin series mintage strike & luster data, etc. and feed into a program, and applied quality lighting & photography, AI could easily apply such software to quality & consistent grading in the future.

    Now why should the best graders cooperate to work on one system? Unless there is going to be one TPGS.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @mr1931S said:

    @jmlanzaf said:

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Substitute TPGS for robot and you can ask the same question.

    Slabbing makes the product more marketable so that's in the best interests of the dealer trying to find a buyer.

    Non sequitur. Has nothing to do with what I said. Are you saying that you would buy a slabbed coin that you don't like?

    But, to your non sequitur, for all you know, robot slabbing might also create a more marketable product.

    It is late. I'm a HS dropout and I don't have a dictionary. What is "non sequitur."

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Looks like another decade of work is needed for the AI to be programed and start learning. And...if no one uses the new service the learning process will take a very long time.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    Sounds a little like one of my teachers. Use high magnification to see everything there is to see good and bad on a coin and then back off to the "commercial market acceptability level" to finalize the grade. AI will see everything so you better dumb it down!

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    I think coin conservation involves some type of cleaning. I guess they will teach it to know the difference between good cleaning and bad. LOL.

  • Married2CoinsMarried2Coins Posts: 675 ✭✭✭

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

    If the Google self driving software sees a bicycle and draws the incorrect conclusion, you have a corpse on a Tempe Arizona street.

    Try again.

    So you're not intelligent because you never draw incorrect conclusions?

    Your confusing intelligence with omniscience.

    What is omnisience?

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @Married2Coins said:
    @jmlanzaf

    If you have time to make a short list for us. What will AI be able to do for a TPGS:

    1. AI will be able to identify a coin that it has seen before. (So will an image of the coin stored in a database. I know that ANACS weighed each coin an took a picture of it in the beginning.)

    2. AI will be able to detect fake toning? (Doubtful, humans can't although they think they can.)

    3. AI will be able to measure the reflectivity of luster and depth of PL better and more consistent than humans.(I'm sold!)

    4. What else?

    5.

    6.

    AI will be able to HELP with everything. Again, it is a tool. And it is a tool that works 24/7. It doesn't have to be better. If might not have to even be 100% equivalent. It just needs to be close.

    AI MAY be able to better detect fake toning, authenticity, luster, strike, etc. But not by itself. The application has to be developed. A bad implementation could be worse.

    Consider strike. A human looks at the coin and gets an impression based on experience. The AI on the other hand COULD (depends on implementation) look at a billion data points from across the surface and seen from a dozen angles...in seconds. And if properly told what to look for, it will collate all that data and make a reproducible judgment.

    And don't forget that AI is improving exponentially. ChatGPT is more accurate and more functional than it was a year ago. In 5 years or 10 years...

    So when the AI looks at enough coins to have the experience of a professional grader it will tell him/her what they already know? Sounds like a big waste of time to me. I will give you this much, AI will be able to grade coins sometime in the future with results that are better than many average graders.

    I think Flying Al will do a better job.

    That is not what I said. What i said is that the AI COULD see a billion days points on EACH coin.

    Sorry. I think that would be called "micro-grading" in a class. It is discouraged. Perhaps AI should be set at 10 days points to be like a 5X magnifier and closer to a professional grader and market acceptability.

    The developer can do whatever they want. That is the point. They can grade without looking at the coin at all, if you want. You use whatever algorithm gives you the desired precision.

    Sort of like picking a standard and nobody likes it or uses it. I wonder who the developer is going to be. Can't we all just get along...LOL! NUTS, to AI grading. It WILL happen but will not replace humans.

    PCGS will likely be a developer. NGC will likely be a developer. It's a tool. If you are in the grading business, you will develop your tools.

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @Married2Coins said:

    @jmlanzaf said:

    @mr1931S said:

    @jmlanzaf said:

    @mr1931S said:
    Why should a collector who has taken the time to learn to grade coins for himself or herself give a ra what a robot with no soul has decided his or her coin's grade is?

    Substitute TPGS for robot and you can ask the same question.

    Slabbing makes the product more marketable so that's in the best interests of the dealer trying to find a buyer.

    Non sequitur. Has nothing to do with what I said. Are you saying that you would buy a slabbed coin that you don't like?

    But, to your non sequitur, for all you know, robot slabbing might also create a more marketable product.

    It is late. I'm a HS dropout and I don't have a dictionary. What is "non sequitur."

    It does not follow the point being made. Like if you said, "i love my Tesla roadster" and I replied "I prefer fried chicken to roasted".

  • jmlanzafjmlanzaf Posts: 34,561 ✭✭✭✭✭

    @Married2Coins said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

    If the Google self driving software sees a bicycle and draws the incorrect conclusion, you have a corpse on a Tempe Arizona street.

    Try again.

    So you're not intelligent because you never draw incorrect conclusions?

    Your confusing intelligence with omniscience.

    What is omnisience?

    All knowing.

  • tincuptincup Posts: 5,186 ✭✭✭✭✭

    @jmlanzaf said:

    @Married2Coins said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @DocBenjamin said:

    @jmlanzaf said:

    @messydesk said:

    @coastaljerseyguy said:
    ... AI is being programmed to assist with medical surgeries, I think it can be developed to grade coins.

    For imaging-related medical AI applications, the amount of time and data needed to train, test, and release one application is staggering, and this is starting with equipment that is calibrated to very specific imaging parameters such that the image quality and image impression is predictable and consistent not only from one day to the next on a specific imaging system, but all imaging systems, even across vendors.

    Before AI can grade coins reliably, the image acquisition step needs to be just as refined. The acquisition system needs to be fully automated and collect data that produces the same data that a grader defining the ground truth sees when grading a coin. A single, good 2D photo is often pretty good, but seldom enough. A bad 2D photo is worthless.

    Yes. But this is a solvable problem.

    AI will be used to assist in every area of human life. Anyone who thinks otherwise, really doesn't understand "AI". Even if a human is still the main grader, that human will be using AI tools to help grade coins.

    First, it needs to figure out that cleaning coins is not a good thing.

    It knows whatever you tell it.

    That does not imply intelligence of any dscription.

    Yes it does. If i tell you 10 things and you draw a conclusion, did that require zero "intelligence" on your part?

    If the Google self driving software sees a bicycle and draws the incorrect conclusion, you have a corpse on a Tempe Arizona street.

    Try again.

    So you're not intelligent because you never draw incorrect conclusions?

    Your confusing intelligence with omniscience.

    What is omnisience?

    All knowing.

    Perhaps the correct word/spelling is omniscience?

    ----- kj
  • tincuptincup Posts: 5,186 ✭✭✭✭✭

    The ONLY use that I myself would be interested in AI would be for detecting counterfeits. I would not accept or trust AI graded coins. But then I have never put total trust in certified coins either.

    It appears the art/science of an individual being able to grade coins themselves is becoming one of the lost arts, along with cursive handwriting, being able to perform calculations with out a calculator, studying and learning from history, etc.

    While slabs are useful to me to some extent, I much prefer to use my own judgement. I have just as much interest in a raw low grade coin, as higher grade BU examples.

    ----- kj

Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Emoji
Image
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file