Using AI?

[Paul B4]

How is AI being used to help with coin identification, etc.?
Thank thanks for sharing your AI experiences.

[Joe Sermarini]

I have not tried using AI for coin identification. I try using it often for other purposes, but the "hallucinations" make it difficult.

[Lech Stępniewski]

If facial recognition works so well for a billion Chinese people in China, then it must be possible. However, someone would have to invest a lot of money in it. Maybe in five years, cheap AI will be good enough to make it worthwhile. For now, it's still better to practice on your own.

[Bebek2006]

How is AI being used to help with coin identification, etc.?
Thank thanks for sharing your AI experiences.

Hi,
I tried identifying a coin using AI, but generally it does it very poorly...

[Joe Sermarini]

I have tried to use AI so many times without success. Sometimes I feel like it is lazy. Sometimes it seems like it is angry and doesn't like working for a stupid human. Just now I asked it to reformat a bibliography so it matched my format. It kept making ridiculous mistakes, including inserting Chinese into the text. I asked it to stop adding Chinese, but it did it again and again.

[Paul B4]

Thank you for sharing your experiences with AI.  Looks like it needs development for our use.

Have you gotten good results by using Google image search?

[Pawel B]

Not really — the problem isn’t the technology. The commonly available AI models simply aren’t trained or optimized for ancient coin recognition, so their results are limited. With a good, well-labeled dataset, even simple models will work extremely well.

[Lech Stępniewski]

With a good, well-labeled dataset

As I wrote, someone would have to invest serious money. It seems that training on well-prepared data is the key issue for AI today.

[Meepzorp]

Hi folks,

This is funny.


Meepzorp