Interdisciplinary　research　teams　are　crucial　in　solving　complex　problems　by　providing　creative　solutions　that　single-discipline　teams　cannot　achieve.　Despite　considerable　research　has　been　conducted　to　enhance　the　efficacy　of　interdisciplinary　teams,　there　is　still　a　lack　of　understanding　regarding　the　correlation　between　team　diversity　and　innovative　performance.　Therefore,　this　study　investigates　this　question　thoroughly　with　the　most　influential　scholars　and　their　collaborators　in　artificial　intelligence.　Furthermore,　decision　tree　algorithms　were　utilized　to　examine　which　interdisciplinary　teams　(according　to　diversity　characteristics)　are　more　likely　to　achieve　high　innovation　performance,　measured　by　novelty　and　impact.　The　results　of　the　study　show　a　U-shaped　relationship　between　a　combination　of　research　interests　diversity　and　member　diversity　and　the　＂novelty＂　innovation　performance.　Specifically,　teams　exhibiting　high　diversity　in　research　interests　tend　to　demonstrate　superior　innovative　performance,　irrespective　of　member　diversity.　Conversely,　teams　with　low　research　interest　diversity　can　only　attain　higher　novelty　in　their　innovative　performance　if　member　diversity　surpasses　a　certain　threshold;　otherwise,　their　novelty　performance　diminishes.　Regarding　＂impact＂　innovation　performance,　teams　characterized　by　higher　member　diversity,　while　maintaining　research　interest　diversity　within　a　reasonable　range,　are　likely　to　achieve　higher　impact.　Additionally,　interdisciplinary　teams　that　exhibit　lower　member　diversity　but　higher　institutional　diversity　also　demonstrate　enhanced　performance.　Moreover,　the　study　found　that　research　interest　diversity　served　as　the　variable　most　strongly　associated　with　team　innovation　performance.　This　study　extends　the　research　on　the　complex　non-linear　relationship　between　multi-factor　combinations　of　team　diversity　and　the　innovative　performance　of　interdisciplinary　research　teams.