New Landscape: Deep Learning and Other Breakthroughs
Pietro Perconti, Alessio Plebe · 2025
This chapter introduces the revolution in AI caused by what is commonly referred to as “Deep Learning”, recounting how it has revived a discipline that seemed on the verge of exhaustion. The vicissitudes of AI in pursuing Turing’s dream have for a long time ruthlessly highlighted the gap between dreams and reality. In certain periods, the frustration due to lack of results threatened to transform the dream into real nightmares. The history of AI is marked by periodic moments of this kind, known as “AI winters.” The most regrettable consequences of these winters were the freezing of research funds, which forced many research centers to change course towards other technologies. A memorable winter of this kind followed the investigation commissioned in 1971 by the UK Science Research Council, the primary funder of AI research in the UK, to mathematician Sir James Lighthill. His report was extremely negative, highlighting the dissatisfaction with the results achieved after twenty years of research. Unlike literal winters, this cold season for AI lasted for over a decade. After a gradual recovery, starting from the 80s, the beginning of this millennium was again a tough time for AI. In particular, Turing’s vision to realize human-like intelligence clashed with frustrating failures. Then came Deep Learning. Its arrival was completely unexpected, and its disruptive impact lifted AI from one of its most lethargic moments to a credibility it had never experienced before. For good reasons, concrete realizations of cognitive capabilities directly comparable with human ones were finally glimpsed.