OpenAI and Los Alamos National Laboratory Announce Research Partnership

OpenAI and Los Alamos National Laboratory (LANL) have entered into a research partnership to safely leverage artificial intelligence in bioscience research. This collaboration is part of an effort to evaluate the biological capabilities of frontier AI models and mitigate related risks, in accordance with the White House's executive order.

Both parties will conduct research evaluating how multimodal models such as GPT-4o can assist scientists in real laboratory settings through their visual and voice capabilities. The goal is to maximize the speed and impact of scientific research while preventing the risk of AI being misused for biological threats.

This evaluation will be conducted by measuring the ability of both experts and novices to carry out standard experimental protocols and solve problems. The key experimental tasks include the following:

• Transfection (introduction of foreign genetic material)
• Cell culture (maintaining and growing cells in vitro)
• Cell separation (e.g., centrifugation)

Notably, this research goes beyond existing text-based evaluations by combining actual wet lab techniques with multimodal capabilities. Through this, the plan is to quantitatively analyze how much AI can improve the accuracy and proficiency of actual laboratory work.