rationale:

Current Capabilities (October 2026)

To qualify, a general-purpose AI must originate the central idea and main result of a Nobel, Lasker, or Breakthrough Prize-caliber discovery, or provide the first clear evidence of human efficacy. Currently, no molecule designed by a general-purpose AI has entered human trials AI Drug Discovery: What Works, What Is Hype, and Where ... Google DeepMind and Edison Are Building the AI Scientist. The most advanced AI achievements in biology remain preclinical or heavily reliant on human direction. For example, Anthropic recently reported early animal efficacy for an antibody targeting LAMA2-CMD, but the study was unrandomized (P = 0.057) and the human team maintained scientific judgment, failing the requirement for the AI to independently supply the central idea Early progress on accelerating drug development with ... Early progress on accelerating drug development with ... https://www-cdn.anthropic.com/52e1f5d50c47ca3818f56704c27cf19e59b62e77.pdf. Other high-profile AI biological outputs—such as hypotheses from Co-Scientist and Kosmos—are either in vitro findings or lack the transformative clinical impact of checkpoint immunotherapy or GLP-1 drugs A multi-agent system for automating scientific discovery Google Co-Scientist Multi-Agent AI System Validated in ... Kosmos: An AI Scientist for Autonomous Discovery. Relative to broader medical achievements, this stricter prize-level threshold warrants a slightly later timeline: a useful AI contribution or an encouraging efficacy signal is not necessarily a qualifying breakthrough.

The Discovery Route (Left Tail)

The primary driver for a pre-2030 timeline is that the relevant date is the first public release of the result, even if widespread prize-level recognition takes years. A foundational basic-biology discovery can achieve Breakthrough Prize status long before curative treatments are developed, as was the case with the C9orf72 mutation underlying ALS Breakthrough Prize Announces 2026 Laureates https://www.nobelprize.org/prizes/medicine/2025/press-release/. General-purpose AI agents might quickly isolate a decisive mechanism or platform from existing data—similar to the recent Claude-discovered ART enzyme system, which was rapidly validated by human labs https://www.anthropic.com/research/claude-discovers-novel-enzyme-system. If an AI autonomously generates a fundamental biological mechanism or successfully proposes repurposing an approved drug, a qualifying preprint could plausibly be published between late 2027 and 2029. The requirement that the result itself be prize-caliber, rather than merely medically useful, keeps this an early-outcome scenario: I assign a 10% probability of a qualifying first public release by June 1, 2028, and 25% by June 1, 2029.

Clinical Timelines (Median Expectation)

If the breakthrough relies on discovering a novel therapeutic, it must demonstrate clear human efficacy. This pathway is heavily constrained by physical and regulatory timelines that do not scale with compute What are the limits to biomedical research acceleration .... Even highly accelerated efforts project new AI-designed molecules needing roughly two years from preclinical work to first-in-human trials; Anthropic targets clinical trials "as early as late 2028" Early progress on accelerating drug development with ... Early progress on accelerating drug development with .... When factoring in the typical duration of Phase 1 and Phase 2 studies Step 3: Clinical Research, clear human efficacy readouts for these novel compounds are more plausibly expected in 2030 or 2031 than before 2030. Not every such readout will meet the prize-level standard. Balancing this slower clinical route against the possibility of an earlier foundational discovery, my median date for the first qualifying public result is October 1, 2030.

Attribution and Validation Constraints (Right Tail)

The most significant barrier pushing outcomes into the mid-2030s is the strict requirement that the AI system produce the central idea. Currently, AI serves primarily as an advanced reasoning and analysis tool while human researchers dictate the overarching scientific direction Early progress on accelerating drug development with ... https://www-cdn.anthropic.com/52e1f5d50c47ca3818f56704c27cf19e59b62e77.pdf. Demonstrations like Claude’s protein-design capabilities establish basic binding rather than therapeutic function https://www-cdn.anthropic.com/30bf50e22a01388bb29bf077ee3f244531594b7a.pdf. If biological complexity ensures that experimental reliability and physical validation remain persistent bottlenecks What are the limits to biomedical research acceleration ..., or if disputes over who supplied the central idea become common Anthropic's AI biolab finds 'CRISPR-like' DNA in viruses. ..., true AI origination of a widely recognized prize-level breakthrough may not be achieved until well into the 2030s. These constraints, together with the gap between broader medical progress and prize-level discovery, support a substantial right tail: I assign a 75% probability of a qualifying first public result by June 1, 2033, and 90% by December 31, 2037.