TigerAI: An AI-powered genetic evidence platform to support clinical development

Genetic evidence plays a critical role in determining the success of drug development, yet its synthesis has traditionally been a time-consuming and labor-intensive human process. Large language models (LLMs) present a promising avenue to rapidly generate genetic evidence through automated synthesis of biomedical knowledge. To assess the capabilities of these models, researchers developed a specialized framework to evaluate GPT-5, the latest generative language model, in generating genetic evidence relevant to clinical trial outcomes.

By benchmarking GPT-5 against 13,022 target-indication pairs sourced from a comprehensive drug development database, the study compared the AI-generated evidence to a recently completed human expert-curated study. The findings revealed that GPT-5 produced genetic evidence that was at least as informative as the expert-curated data for inferring clinical success, while significantly broadening the scope of available evidence beyond traditional curation resources.

Leveraging these results, the researchers unveiled TigerAI, a dual-purpose platform designed to assess the performance of cutting-edge LLMs and serve as an accessible resource for obtaining trustworthy AI-generated genetic evidence. By integrating AI-powered genetic evidence into drug development pipelines, these innovations harness the potential of LLMs to enhance the likelihood of clinical success.