Abstract: [Purpose/significance] In the new research environment, research achievements are available on diverse formats and show different impacts. This paper compares impact of books and papers in artificial intelligence from various angles, and provides reference for comprehensive academic evaluation in the new environment. [Method/process] The paper combines bibliometrics and Altmetrics, and selects academic pa-pers and books in artificial intelligence, then compares them from aspects of country/author/age distribution, impact indicators, and subject content. [Result/Conclusion] In the field of artificial intelligence, for the number of published papers and books, the distributions of countries are relatively similar; the high-yield authors of both formats overlap little. In general, whether citation or altmetric indicators, the values of these indicators in papers are higher than those in books. In the aspect of subject, In the aspect of subject, topic words of papers are more novel and specific, reflecting the current research hotspots; but those of books are more general and explore the hotspots in the ascending stage. Both are important to analyse development stage of the field. [Limitations] This paper only focused on the field of artificial intelligence. In the future, we can expand the scope of research, obtain more data, combine more altmetric indicators to evaluate the impact of academic achievements more comprehensively.

Key words: artificial intelligence, books, bibliometrics, Altmetrics, impact evaluation