AI models need more data about biology, and OpenAI is paying to create it (www.technologyreview.com)

OpenAI Foundation has announced funding to enhance AI's role in medicine by creating high-quality scientific datasets, particularly using insights from failed biotech companies. This initiative, known as Public Data for Health, aims to address the current data scarcity that limits AI's capabilities in drug development and approval processes. Ruxandra Teslo's concept of acquiring “biotech’s lost archive” through bankruptcy proceedings will be supported with a $500,000 grant, underscoring the potential for this archival data to serve as a crucial training resource for AI models. The significance of this effort lies in its potential to unlock critical information that can expedite the development of medical treatments and improve the understanding of regulatory challenges. As Morgan Levine notes, access to more comprehensive datasets is essential for AI to contribute effectively to biological breakthroughs. The initiative also includes a $40 million grant for cancer vaccine data collection and supports competitions aimed at predicting drug effects. As OpenAI navigates its dual role of advancing AI technologies while addressing broader societal concerns, the foundation aims to distribute over $1 billion in grants by the year’s end, marking a significant investment in the intersection of AI and healthcare.