• Put GenAI back into the long history of AI.
• Show that each technology has an optimal scope.
• Demonstrate that “doing everything with GenAI” is often:

GenAI fascinates, impresses, and sometimes steamrolls everything in its path. But do we really need a giant LLM burning gigawatts just to extract three lines of text from a PDF?
This talk retraces the evolution of AI technologies — from OCR to classical models, all the way to GenAI — to show how a reasoned, sustainable, and modular approach can reduce carbon footprint, costs, and even technical risks.
Spoiler: the “smartest” solution isn’t always the biggest one.