Teams now drive most scientific advances, yet the impact of absolute beginners -- authors with no prior publications -- remains understudied. Analyzing over 29 million articles published between 1941 and 2020 across disciplines and team sizes, we uncover a near-universal and previously undocumented pattern: teams with a higher fraction of beginners are systematically more disruptive and innovative. Their contributions are linked to distinct knowledge-integration behaviors, including drawing on broader and less canonical prior work and producing more atypical recombinations. Collaboration structure further shapes outcomes: disruption is high when beginners work with early-career colleagues or with co-authors who have disruptive track records. Although disruption and citations are negatively correlated overall, highly disruptive papers from beginner-heavy teams are highly cited. These findings reveal a ``beginner's charm'' in science, highlighting the underrecognized yet powerful value of beginner fractions in teams and suggesting actionable strategies for fostering a thriving ecosystem of innovation in science and technology.