The OpenAI Foundation said Tuesday it is launching Data for Public Health \u2014 a grant push to build \u201chigh-quality scientific datasets\u201d so medical AI has more real biology to learn from, MIT Technology Review reported.<\/p>\n
In the first round, the foundation said it would give $40 million to a University of North Carolina\u2013Chapel Hill program collecting data on novel cancer vaccines, and will also support OpenAdmet, which runs competitions where researchers try to predict drug effects. A separate $500,000 grant goes to 1Day Sooner to pursue analyst Ruxandra Teslo\u2019s idea of buying \u201cbiotech\u2019s lost archive\u201d \u2014 regulatory filings and safety data from failed biotech companies that usually stay locked as trade secrets.<\/p>\n
\u201cEveryone is recognizing that data is the biggest bottleneck in successfully applying AI to biology,\u201d former Altos Labs computation VP Morgan Levine told the outlet. The foundation said remaining breakthroughs in preventing and curing disease will come from pairing new models with more observations of the world \u2014 more data.<\/p>\n
OpenAI\u2019s nonprofit parent holds a large equity stake in the for-profit lab and has ramped grantmaking this year, including a prior $100 million gift for hepatitis C access. Foundation executive Jacob Trefethen said the charity operates separately from the product company and hopes to give away $1 billion by year-end.<\/p>\n