I asked ChatGPT…

I asked ChatGPT to translate from Italian into English the sentence “l’importanza di un database come Nomismata” and GPT replied:

The importance of a database like **Nomismata** lies in its ability to systematically collect, organize, and make accessible a vast amount of numismatic data. Such a database provides scholars, collectors, and researchers with reliable and structured information on coins, including their historical context, typology, and provenance. By centralizing this information, **Nomismata** facilitates comparative studies, enhances historical research, and supports the authentication and valuation of coins. In addition, it contributes to the preservation of numismatic heritage by ensuring that valuable data is not lost over time. In a digital and interconnected world, a specialized database like **Nomismata** is an indispensable tool for advancing numismatic studies and promoting cultural heritage.

I confess that I was stunned. How does GPT know about Nomismata? I simply wanted to translate from Italian the phrase “the importance of a database like Nomismata” and GPT replied as if it understood “the importance of Nomismata.” But how did GPT do it? The answer is simple. I used GPT to translate this entire site from Italian to English. And in doing so GPT “learned” what Nomismata is, what its importance is, etc.

Every time we ask the AI for something, without realizing it we “train” it. Disturbing.

Actually these days I am experimenting with applying AI to Nomismata. Particularly for creating templates to feed the database from sources (books, articles). Again, the result is both amazing and promising at the same time. If we feed GPT (or any generative artificial intelligence model) a good-quality pdf and tell it “generate an Excel file by putting denomination, authority, date of issue, catalog number, mint,…” in the vast majority of cases it does so. We thus get a more or less half-finished template in a few minutes, depending on what we were able to get the model to do. I am only in the early days of these experiments, but what I have seen so far is very promising. I will keep you updated.

Ciao

Giorgio