There is a long history of applications of various Artificial Intelligence (AI) technologies in software engineering. From machine learning, evolutionary computing, to Natural Language Processing, AI has played an increasingly important role in making software engineering more predictable and automatable. This rising impact stems from increasingly powerful AI technologies, easy access to enormous computing power, and the availability of large amounts of software data in readily available development repositories. This talk will provide a reflection over 25 years of experience in applying and tailoring AI techniques to address software engineering problems at scale. I will try to abstract away and identify patterns of hard software engineering problems requiring solutions based on AI. Challenges will be characterized and interdisciplinary research avenues will be outlined.