OpenAI has expanded its research focus into longevity science by developing an AI model to enhance the efficiency of stem cell manufacturing. The model GPT-4b micro, is designed to aid in stem cell production. This model focuses on optimizing Yamanaka factors—proteins used to reprogram regular cells into stem cells.
The collaboration with Retro Biosciences, a San Francisco-based company working to extend the human lifespan, is part of OpenAI’s first major step into biological research.
The reprogramming of cells using Yamanaka factors typically takes weeks, and fewer than 1% of cells treated in the lab are successfully transformed into stem cells. What distinguishes GPT-4b micro from other AI models, such as Google’s AlphaFold, is its focus on protein interaction rather than structure prediction. The model was trained using protein sequences from different species and aimed to understand how various proteins interact with one another.
In practice, Retro scientists used the model’s suggestions to modify the proteins, which, in early tests, reportedly made them over 50 times more effective. According to OpenAI, these modifications appear to outperform the results that scientists were able to achieve independently.
The potential impact of this model is significant for longevity and regenerative medicine. Researchers, like Harvard’s Vadim Gladyshev, believe that more efficient methods of creating stem cells could be instrumental in treating aging-related diseases and even developing new tissue or organs.
