EN
AIn a search for lessons learned from 50 years of history of AI, this paper presents a brief, subjective and personal history of the field. It then introduces five theses-prescriptions for what makes good AI research. The theses stem form the author's understanding of successes and failures of the field, and from his own experience as a long-standing and active member of the AI community. The five theses promote practicality, embeddedness, empirical verification, mathematical foundation, and scrutiny.