Claude did not invent scorpion antivenom (www.theverge.com)

Citrini Research, in partnership with Anthropic, attempted to leverage AI, specifically an unreleased version of Claude, to develop a new scorpion antivenom. Although the AI generated nearly 320,000 protein candidates designed to target scorpion toxins, the results revealed that while some candidates showed promise, not a single protein was able to bind effectively to more than one toxin. Experts highlighted that while the initial discovery of binding proteins is a relatively achievable step, the more significant challenge lies in translating these findings into a safe, effective medication, which the Citrini team acknowledged has not yet been attempted. The significance of this experiment for the AI/ML community rests in its exploration of the autonomous drug discovery process. While Citrini's findings demonstrate the potential for AI to accelerate the early stages of drug development, experts emphasize that substantial hurdles remain in validating these proteins for therapeutic use. This venture illustrates a growing trend of integrating AI into life sciences, underlining both the possibilities and limitations of using technology for complex biological challenges, as well as signaling investment opportunities in AI-enabled drug discovery.