AI just broke through a major bottleneck in coin analysis

My work on AI coin analysis and grading has me closely following advancements in AI models, and the industry is evolving rapidly.

Previously, AI could only analyze static images of coins, which severely limited accuracy.



Collectors know the importance of viewing coins from multiple angles and lighting. Well...Google’s just updated Gemini 1.5 Pro to support video uploads. It can now analyze thousands of frames instead of just a few images. This is a game-changer.

I recorded a short video to check whether it would even return coherent results. I ran a quick test on a slabbed 1925 Lincoln Cent. The AI accurately identified the coin, though the grade was slightly off. That's okay, though. If I had used a high-resolution camera, set up better lighting, conducted additional tests to determine the optimal video length, and fine-tuned the model, I bet I would have gotten an accurate grade.

That's not to mention that OpenAI just released the best AI model in the market, the o1 model, which now has PhD level intelligence and reasoning. That model can't take video inputs yet, but I'm excited to test it out when it does.

In short, AI can now mimic the conditions of a human grader.

 

Log in or Sign up to hide this ad.

No. Just.....well.... no. It can not.

I also saw that movie. They went around killing everybody.

So.....

 

Perhaps, but I just got AI-generated spam from a gambling site advising me to bet on the "Eco-friendly Bay Packers" so call me skeptical, at least for now.

 

Are people really that lazy that they want a computer to grade their coins? The best grader I know are my own eyes.

 

Not to belittle the ability of your eyesight in any manner or suggest a lack of capability of any quality, my favorite eyes are those of JA.

 

Judging by the garbage coin descriptions ebay's AI bot generates, I'd say accurate grading is still a long ways off. If it does become a thing, I sure hope there's a disclaimer attached to anything graded with it. Frankly I find the whole thing insidious. People already have enough trouble knowing what is real and what isn't, and we're being pushed into a world where nothing is real? No thanks.

 

Couldn't agree more strongly. We WILL pay the price for all this nonsense in the near future.

Bruce

 

Sounds like you want a "near me but in the future" option, which makes such total sense that it seems stupid that no one has thought of that.

Seems hard to imagine AI ever telling the difference between a weak strike and wear, or a rusted die from environmental damage, etc etc. But fundamentally, when AI is doing everything, how is anybody supposed to pay for anything? Even the jobs running the computers and fixing the robots will be done by AI and robots. Humans become irrelevant, which is already covered by quite a few books and movies...

 

While the intellect can identify things and call things by their proper name, it can’t appreciate the meaning or significance of those things. It takes emotions for that. That’s the reason God gave us emotions. And that’s why AI will never replace us. Simply put, it hasn’t a clue. It’s like a 4-year-old who has the intellectual capacity to identify the fact its pet died, that’s it, that’s all the good it is, it hasn’t the first idea what death is. And that’s what we’re going to let make decisions for us? What are we, cracked? We’re going to let a 4-year-old who can’t judge the meaning or significance of anything dictate to all us? It’s as dead as that pet and is as utterly lost as that 4-year-old child in terms of its emotional development. Did we lose sight of that? I think we did, big time. It comes to this tech calling the shots for us, we’re cracked. This whole cockeyed world is.

 

AGREED!

Bruce

 

Great discussion everyone. Some thoughts to share after I've had time to think through where all this is heading.

I shared my thoughts on last week's The E-Sylum, I would append my thoughts on where AI grading is headed below:

"In short, AI is now just two steps away from surpassing human graders. The next step is AI being able to analyze a live video feed. The final step is integrating live video into a robotics platform. The AI would be able to analyze a coin in real-time. Like a human, it can dynamically rotate a coin to see all angles. But beyond a human, it can:

• Adjust to any lighting conditions.
• Quickly snap from 2x to 20x lens to inspect specific spots.
• Switch to different viewing modes like LiDAR for 3D scanning.
• X-rays are used to determine the composition of counterfeits and identify them.
• And so much more

At some point the narrative will shift from "AI can now mimic a human grader" to "AI is better than a human grader in every way".

I will note that my underlying assumption is AI using technical grading standards. Market grading is a whole other beast."

The biggest issue is that a lot the ways that AI is being implemented is for specific use cases. Gmail can summarize emails. Google docs can reword your documents. But so far, most of these actions aren't connected...yet. I know that Google and others are working feverishly to get AI's to effectively tie all these things together. One path is through 'agentic swarms'. Basically, agents that you can tell to do a certain action, and it will go out and do them.

If one of your turning tests isn't agentic but more on its raw reasoning and intellectual power, OpenAI's o1 model may pass the test. You can test the model out yourself if you have an OpenAI subscription. If not, and you have a question you want to ask, feel free to let me know, and I will post my results. Keep in mind that this model can't analyze images yet.

This is a big outstanding question I have as well. Compugrade struggled with weak strikes and wear. I suspect this will be overcome with live video analysis and real-time lighting adjustments.

 

You can’t pre-program emotions into it. That’s what you’re not understanding. You can’t program it to weigh luster or strike against contact marks and come up with a grade, you can only get it to identify those attributes and “grade” by pre-programmed, arbitrarily-assigned “weights,” as though the grade were but a purely intellectual construct. Teach AI how to evaluate toning. A contact mark under the eye vs. one in the hair vs. in the field. Coins are art and there are pros and cons on every one that affect the way we feel about the coin, and AI can’t feel, there’s the bottom line. Every coin engraver worth anything is skillfully employing the elements of his craft to consciously control our eye-movement through the coin, which, in turn, is the chief determinant in how we come away feeling about the coin. We’re the market in this market grading game. I’m going to let some mathematician quantify that activity for me? Maybe you are, I’m not. That’s not how we grade our coins. It’s certainly not how we market grade the coins. We’re not pre-programmed for outcomes, like we’re but machines. We evaluate the strengths and weakness of the grade through feelings, not through predetermined mathematical formulae, or something, namely, as this…

 

I don't believe AI grading would be horrible. I actually don't think it would be any different than submitting a coin to any major TPG company. Why, you ask? Good question. Let me explain.

Grading isn't a science. It isn't mathematical.
We've all seen a holdered coin in a slab that we don't agree with the grade. That's why we see people resubmit a coin in an NGC holder to PCGS for regrading. Or we see a coin in an ANACS holder and say it can do better in an NGC holder. And these are coins graded by humans. If humans can't agree on a grade, unanimously, how can a computer do it where everyone agrees unanimously?

Another example. The existence of CGC. They base their entire business model on the fact that people don't agree on grading. They make millions of dollars confirming, with green beans, what someone else already did. How about the coveted gold bean? Coins that have a grade below what CGC 'dictates' is the appropriate grade.

Either way, CGC exists because people don't agree on a grade. Why would an AI computer grade be any less subjective?

 

But if the grade assigned to a coin depends on how the grader feels at that particular moment... how is that a good thing?

How I feel about a particular coin in seeing for the first time can't possibly be independent of how I feel in general at that moment. And how I feel in general is completely irrelevant to the coin's condition or quality.

But, hey, without inconsistent grading, the crack-out game goes away, along with a big chunk of TPG business. And we all know how losing business feels.

 

A recent WSJ article...

https://www.wsj.com/tech/ai/yann-lecun-ai-meta-aa59e2f5

I don't have a subscription but the headline reads...

This AI Pioneer Thinks AI Is Dumber Than a Cat
Yann LeCun, an NYU professor and senior researcher at Meta Platforms, says warnings about the technology’s existential peril are ‘complete B.S.’

 

But if someone is expert enough to do video and lighting adjustments specifically tailored to grading a coin, they could just grade the coin themselves instead of going through all that just to pump it into AI.

 

If someone knows how to, then yes, they could grade themselves. However, collectors who aren't experts would have access to accurate grades without spending a lot of money or building up years of experience.