The Stanford study also confirmed that open source models are much more transparent than the more popular—and profitable—models like ChatGPT and Bard.