OpenAI: GPT-5 Demonstrates the Ability to Assist Scientific Research in Real Laboratory Environments

According to AXIOS, OpenAI has revealed for the first time that GPT-5 has demonstrated the ability to assist scientific research in real laboratory environments, indicating that artificial intelligence is poised to play a more central role in scientific experiments. While AI has made rapid advances in fields like mathematics and physics, significant breakthroughs in biology have been relatively slow, as biology relies heavily on real-world laboratory work rather than just computational simulations. OpenAI partnered with the biosafety startup Red Queen Bio to build a testing framework to evaluate AI models' performance in the laboratory. The laboratory involves 'wet' operations such as liquids, chemicals, and biological samples, distinguishing it from 'dry' laboratories focused on data analysis. In the experiments, GPT-5 proposed improvements to the molecular cloning experimental process, which human scientists executed, providing feedback to GPT-5, which iteratively optimized the proposals based on the feedback. Experimental results showed that GPT-5 improved the efficiency of a standard molecular cloning process by 79 times. So amazing!