The first stage, then, was the creation of an open and accessible database. The second stage, currently underway, is transforming the database into an advanced scientific repository. To this end, thousands of archives, laboratory data, and additional information that the IAA collected are being fed into it, and in the future it will include research articles from around the world.The first stage is done. Great. AI will be quite useful for the second stage, which amounts to advanced bean counting.“In the second stage, which we are in right now,” said [Alby] Malka, head of the IAA Technologies Division], “the Israel Antiquities Authority is using Google’s artificial intelligence tools. A pilot project has already scanned 5,000 archaeological research articles relevant to finds discovered in Israel.”
The third and most ambitious stage will enable the AI system not only to retrieve information but also make connections, suggest new perspectives, and answer complex questions. The system will integrate archaeological finds, scientific articles, and general knowledge, performing a synthesis that can serve as a research tool. Completing this stage will undoubtedly constitute a significant breakthrough in combining new technology with the findings of the past.
The catch comes with the third stage. The achilles heel of LLM AI is that there is no one home. Its a collection of automated processes—advanced autocorrect—which give a result with no one reviewing it against baseline reality. That's why it hallucinates. AI will speed up the third stage as long as humans oversee it to correct its misapprehensions.
When I get a chance, I may say more about this. Meanwhile, this is a good article that understands the benefits and limitations of AI.
Cross-file under Technology Watch.
Visit PaleoJudaica daily for the latest news on ancient Judaism and the biblical world.
