OpenAI's GPT-4 Poses Minimal Risk in Aiding Biological Threat Development, Study Finds

Despite the low risk, the company emphasizes the need for continued research and vigilance as AI technology evolves.

OpenAI's most advanced AI model, GPT-4, poses a minimal risk of aiding in the development of biological threats, according to a study conducted by the company.
The study involved 100 participants, including 50 biology experts and 50 students, who were tasked with completing a set of tasks related to biological threat creation.
Participants were divided into two groups, one with access to GPT-4 and the internet, and the other with access only to the internet.
While the group using GPT-4 showed a slight increase in accuracy and completeness of tasks, the results were not statistically significant.
Despite the minimal risk posed by GPT-4, OpenAI emphasizes the need for continued research and vigilance as AI technology continues to evolve.