An end-of-year experiment exploring how large language models perceive leadership in AI-driven longevity.
As the year draws to a close, it feels increasingly difficult to ignore how much large language models now shape what we see, read and remember. They summarize research, surface experts, rank relevance and, increasingly, act as quiet arbiters of authority across fields that remain fluid, contested and fast-moving. Longevity science – and particularly its intersection with artificial intelligence – sits squarely in that category.
So, rather than offering a traditional editorial ranking, Longevity.Technology set up a small experiment. We asked multiple large language models the same structured question: who they identify as the single leading individual worldwide in the application of artificial intelligence to human longevity and aging research, and who they consider the next tier of leaders. The aim was not to outsource judgment, but to examine what patterns emerge when AI systems are asked to synthesize leadership in their own domain of influence.
How the experiment worked
The same prompt was posed independently to several widely used LLMs, including ChatGPT, Gemini, Grok, DeepSeek, Claude and Perplexity. Each model was asked to identify a ranked list of individuals, alongside brief explanations referencing AI methodology, translational impact, biomarkers, therapeutics and influence on the field. The wording was held constant. Responses were collected without iteration, nudging or follow-up.
What followed was less a vote than a convergence exercise. We were not looking for unanimity, but for consistency: where different models, trained on different corpora and optimized for different goals, nonetheless arrived at similar conclusions.
A striking point of agreement
Despite variations in emphasis and framing, one name appeared with remarkable consistency across every model tested. In each case, the reasoning followed a similar arc: not simply publication volume or visibility, but the sustained application of AI across the full longevity pipeline – from biomarker development and target discovery through to molecule design and clinical translation.
Across models, Alex Zhavoronkov was repeatedly described as standing out for building and operationalizing end-to-end AI platforms explicitly focused on aging biology, and for translating those systems into real therapeutic programs. References to generative chemistry, aging clocks, integrated target-to-clinic workflows and human clinical trials recurred again and again, regardless of the model’s broader tendencies.
Several models emphasized speed and execution, pointing to AI-discovered targets and AI-designed molecules progressing into Phase II trials as evidence that artificial intelligence, when tightly coupled to biology, can move beyond theoretical promise. Others highlighted infrastructure and influence, noting the extent to which these platforms have been adopted by pharmaceutical partners and shaped expectations of what “AI-first” drug discovery looks like in practice.
Zhavoronkov himself was cautious about how much weight to place on such an exercise. Reflecting on the results, he told us: “When I first saw that every LLM ranks me #1 in AI for longevity, I was quite pleased, since when it comes to overall global rankings I would trust LLMs more than industry analysts or media who often prioritize the amount of money raised or affiliations with flashy institutions or investors over overall productivity. On the other hand, it is sad that we still don’t have a single AI-discovered longevity drug approved. Delivering novel efficacious longevity therapeutics should be industry’s main objective.”
The comment underlines both the appeal and the limits of the experiment; recognition by pattern-seeking systems is not the same as success measured in patients.
A more fragmented second tier
Beyond the top position, consensus weakened noticeably. While several names recurred, their ordering varied depending on whether a model weighted foundational biology, translational biotech, biomarker development or public influence more heavily.
David Sinclair appeared frequently among the highest-ranked runners-up, particularly in models that emphasized biomarkers, epigenetic clocks and conceptual frameworks that underpin AI applications to aging. His influence on public discourse and on the normalization of biological age as a measurable quantity clearly registers with language models trained on a decade of longevity literature and media.
Kristen Fortney also emerged repeatedly, especially in responses that prioritized human-first data, machine-learning analysis of biobanks and the translation of AI-derived targets into clinical programs. Models tended to frame her work as a counterpoint to more disease-first approaches, emphasizing longevity-specific datasets and outcomes.
Other models elevated figures whose contributions are less about algorithms per se and more about enabling the ecosystem: leaders who have built institutions, standardized datasets or created the biological and clinical infrastructure on which AI methods depend. Here, names such as Vadim Gladyshev, Brian Kennedy and Steve Horvath surfaced, often accompanied by explanations that positioned biomarkers and systems-level biology as prerequisites for any meaningful AI-driven intervention in aging.
Honorable mentions and enabling figures
Beyond the core top five, a wider group of individuals appeared intermittently across responses. Some were recognized for building the massive, multimodal datasets that AI systems require to function at scale; others for advancing clinical frameworks that allow aging to be studied as a modifiable process rather than an abstract risk factor.
Figures such as Craig Venter, Nir Barzilai, Daphne Koller and James Kirkland were cited not because they sit squarely at the center of AI-for-longevity, but because they have shaped adjacent domains – genomics, clinical geroscience, machine learning in biology – in ways that make the current moment possible. In several responses, their inclusion came with explicit caveats, reflecting how LLMs themselves struggle to separate direct application from foundational influence.
What this does – and does not – tell us
It is tempting to read too much into a ranked list, particularly when the results align so cleanly at the top. But this exercise is best understood as a mirror rather than a verdict. Large language models reward coherence, density of signal and demonstrable translation; they surface individuals whose work consistently appears at the intersection of AI methods, aging biology and real-world execution.
That may say as much about how leadership is currently constructed in longevity as it does about who “deserves” it. Models trained on public discourse inevitably reflect where attention, capital and narrative momentum have already converged. They are adept at synthesizing consensus; they are less equipped to identify what has not yet been widely seen.
Still, as AI systems increasingly mediate discovery, funding and reputation, it feels worth paying attention to how they perceive the field they now help shape. If nothing else, the experiment highlights a shift away from abstract promise toward measurable delivery – a signal that longevity science, and its AI ambitions, are being judged less by aspiration and more by execution.
Whether that is reassuring or unsettling may depend on one’s tolerance for algorithms reflecting our collective focus back at us. Either way, it is a conversation that is unlikely to fade in the year ahead.
