AI models show rapid progress in selecting experimental steps

An evaluation by P-Zero Research indicates that AI models are improving quickly at choosing effective experimental steps during research tasks. Top models surpassed human test participants in execution efficiency, although none introduced completely new methodologies.