Martin Shkreli says no and “AI does not help drug discovery”
Resolution criteria
This market will resolve YES if, on or before October 14, 2036, a major regulatory authority (such as the US FDA or European Medicines Agency) grants full regulatory approval for human use to a drug whose molecular design, candidate selection, or target optimization relied primarily on artificial intelligence or machine learning algorithms.
Otherwise, this market will resolve NO.
Notes and Edge Cases:
Approval standard: The drug must receive full regulatory approval for marketing and patient treatment. Clinical trial progress, pre-clinical results, or emergency use authorizations that do not convert to full approval within the timeframe will not trigger a YES resolution.
Role of AI: AI/ML must have played a primary role in discovering, generating, or designing the chemical/biological entity. Standard use of software for clinical trial logistics, administrative work, or conventional data processing does not qualify.
Definition of "relevant": Full approval by a recognized health authority for human therapeutic use satisfies the relevance condition for this market.
Background
Former pharmaceutical executive Martin Shkreli has publicly skepticism about AI in biotechnology, asserting that AI does not meaningfully assist in drug discovery due to the complex biological realities of human clinical trials.
While AI models have accelerated early-stage target identification and candidate design—shortening preclinical phases for several biotechnology firms—the vast majority of drug candidates traditionally fail during human trials due to toxicity or efficacy issues. As of late 2026, multiple AI-designed therapeutics are advancing through multi-phase clinical trials, but full regulatory approval remains the primary milestone demonstrating commercial and scientific success.
This description was generated by AI. Review and verify everything here yourself. You can edit, replace, or delete any part of this description, including the resolution criteria. You do not need to trust the AI output.
1,000