What the research found

A convening of researchers, founders and clinicians identified critical gaps between AI-driven target discovery and actual therapeutic development in ageing research. The core issue isn't computational power—it's the quality and breadth of underlying biological data. Existing genetic databases contain strong associations with disease (particularly cardiovascular outcomes and smoking), but sparse direct evidence about ageing mechanisms themselves. Meanwhile, foundational datasets on protein-DNA interactions have lost funding and remain limited to a small number of cell lines, constraining what AI models can reliably learn from.

The conversation also exposed uncertainty about how to measure success in longevity trials. Because waiting for mortality data is impractical, researchers have adopted proxies like epigenetic clocks, grip strength and frailty scores. Yet the field lacks consensus on whether shifts in these biomarkers actually predict meaningful improvements in function or lifespan—a distinction regulators take seriously. Participants noted that the bottleneck in drug development has shifted: target identification is now the easier part; the harder problems are trial design, toxicity prediction, and persuading regulators that a biomarker change translates to clinical benefit.

Why it matters for you

If you're tracking epigenetic age or grip strength as markers of intervention success, understand that these are still proxies. Moving an epigenetic clock or improving your 6-minute walk test shows something is happening, but the field hasn't yet established what those changes reliably predict about your actual healthspan or mortality risk. This doesn't mean the measurements are useless—they're sensitive to physiological state—but interpret them as directional signals rather than proof of life extension.

The recognition that ageing may not be a single target also reframes how you should think about interventions. Rather than chasing one "ageing drug," the more pragmatic approach involves addressing specific age-related conditions (cardiovascular disease, muscle loss, metabolic dysfunction) where causal pathways are clearer and clinical benefit easier to demonstrate. A peptide or TRT protocol that improves body composition and metabolic biomarkers in your tracking is building on firmer evidence than one marketed as "slowing ageing" itself.

Finally, the emphasis on trial design matters to you as a self-tracker. The best longevity interventions will likely come from repurposed, well-studied compounds tested in rigorous trials with defined populations—not from speculative combinations of novel agents. This argues for patience: focusing on interventions with human trial data (even if not specifically for longevity) may be more rational than adopting cutting-edge experimental approaches before their effects on actual outcomes are understood.

Caveats

  • This is a summary of discussions and opinions from an expert roundtable, not a systematic review or empirical study
  • No new experimental data or effect sizes presented
  • The funding and data gaps described are real constraints, but the proposed solutions (e.g., "ChatGPT for DNA") remain speculative
  • Regulatory and trial design questions remain genuinely open; no consensus exists on how to prove ageing can be therapeutically targeted
  • The observation that ageing "probably isn't one thing" means frameworks for measuring success may need to remain population- and disease-specific rather than universal