AI and the Golem (going-medieval.com)

Historian and essayist Eleanor Janega uses the Jewish golem legend to critique modern AI culture, arguing that large language models have been built on the unpaid, copyrighted work of creatives and now threaten to displace specialists—historians included—despite being technically incapable of real scholarship. Janega frames tech leaders and AI evangelists as a new “priestly class” claiming arcane powers, while creators face economic harm: companies that actually paid for scraped training data would be financially unviable. The golem stories (Prague’s Yossele and the porridge-making tale) become metaphors for machines that must be constantly supervised or will produce runaway, harmful, or useless output. Technically, Janega emphasizes that current models are essentially next-token predictors trained on vast corpora—capable of plausible text but not of genuine analysis, original discovery, paleographic reading, or reliable fact-checking; they hallucinate, consume large resources, and generate low-quality “pseudo-art.” The piece underscores practical implications for the AI/ML community: data provenance and IP/compensation issues, realistic appraisal of model capabilities, the necessity of human oversight and domain expertise, and governance to prevent economic and cultural harm as automated tools create new forms of labor rather than eliminate meaningful human work.