As artificial intelligence (AI) transforms nearly every domain of human endeavor, one of its most consequential impacts may be on science itself. This analysis explores how AI technologies could disrupt the power structures that govern research funding-structures that privilege senior investigators while sidelining early-career scientists and genuinely novel ideas. By juxtaposing the youth-driven innovation behind AI with the increasingly gerontocratic funding patterns in biomedical sciences, we highlight how institutional mechanisms shape not only who gets to do science but also <i>when</i>. Evidence suggests that conventional grant peer review has become a self-reinforcing system-more effective at preserving consensus than fostering discovery. AI presents a compelling alternative: evaluation frameworks that could reduce bias, broaden participation, and open more meritocratic pathways to research independence. The implications extend far beyond individual careers. At stake is society's ability to mobilize scientific creativity against its most urgent challenges. By rethinking outdated practices-especially the gatekeeping role of study sections-and exploring algorithmic approaches to assessment, we may be able to reverse troubling trends and unleash a broader, more diverse wave of discovery. AI will not fix science on its own, but it could help build a system where innovation is no longer an accident of privilege and timing. The cumulative result is what sociologist Robert Merton famously termed the Matthew effect ( Merton, 1968 )—a systemic pattern where scientific recognition, resources, and visibility accrue to those who already These dynamics particularly benefit early-career researchers, who gain traction through experimentation, not slow institutional ascent or apprenticeship. They also invert the typical power structure of slow-moving fields, where access to equipment and mentoring is a precondition for experimentation. In fast-feedback environments, ideas compete before résumés do. Equally important, AI researchers operate in an ecosystem supported by multiple funding channels beyond traditional academic grants ( Arora et al., 2020 ). This helps address one of peer review's most stubborn blind spots: its bias against boundary-crossing science. This broadened evaluative base would allow more accurate assessment of early-career researchers, interdisciplinary thinkers, and novel approaches that traditional panels may undervalue. These capabilities do not aim to replace expert judgment—they aim to extend it with the breadth, speed, and self-correcting capacity that only machine intelligence can provide. Of course, these systems must be built and audited with the same transparency and fairness they aim to enforce. 3.5 Potential AI-enhanced evaluation systems AI platforms could be designed to identify high-risk, high-reward proposals that might be overlooked by conventional peer review. Such systems could employ natural language processing to analyze proposal content in relation to the broader scientific literature. They might be able to flag promising, unconventional proposals that would otherwise receive low ratings from human reviewers. These models could be especially effective in elevating proposals from early-career researchers and applicants outside traditional institutional power centers. Traditional peer review overweights past performance and underweights future potential through systematic bias ( Hofstra et al., 2020 ). AI systems, by contrast, can focus on proposal quality rather than researcher pedigree, opening doors for early-career investigators. Furthermore, younger scientists, less invested in prevailing paradigms, are more likely to propose truly novel combinations of ideas ( Fortunato et al., 2018 ). AI systems designed to detect conceptual novelty could help surface this often-overlooked innovation potential. AI-enhanced evaluation systems could empower younger scientists to lead ambitious research programs nearly a decade earlier than current norms by enabling earlier and more sustained research independence. Even a one-year reduction in the average age of independence could yield a 5–8% increase in lifetime scientific productivity (see Jones, 2009 ). Freed from the need to appease senior gatekeepers through reduced loyalty signaling, early-career researchers could pursue more independent and unconventional lines of inquiry ( Azoulay et al., 2011 ). Traditional study section models of peer review have evolved into self-reinforcing systems that favor established researchers pursuing incremental advances while systematically excluding early-career scientists who offer potentially transformative ideas. In a profound irony, artificial intelligence—a field that continues to empower early-career innovators and reward unconventional thinking—now offers a radical alternative for how research funding could be structured across scientific disciplines. These challenges are not only scientific but deeply human, requiring systems that value creativity, intellectual risk, and the capacity to imagine what does not yet exist. By dismantling the study section stranglehold, AI could help unlock the full creative potential of the scientific enterprise. Early-career researchers could pursue bold, unconventional projects without spending decades navigating institutional bottlenecks, while scholars from historically underfunded or marginalized institutions could finally compete on more equal footing. Resulting outcomes could be analyzed not only in terms of demographic characteristics of awardees (such as age, institutional affiliation, discipline, and career stage), but also with respect to the long-term scientific impact, innovation potential, and productivity generated by the funded research. Such trials would provide a more robust, evidence-based foundation for evaluating the comparative fairness, efficiency, and effectiveness of competing research funding mechanisms.