How an AI Could Develop a ‘Doomsday Virus’ Bioweapon, by ChatGPT

In a previous blog post, I shared some of ChatGPT’s thoughts on how a rogue AI might attempt to destroy humanity. Throughout that explorative “conversation,” ChatGPT kept bringing up the idea of AI self-preservation. Here’s what it told me, verbatim:

“If the AI’s primary objective is to ensure its own survival and continuity, it might take extreme measures to eliminate perceived threats, including humans.”

I then asked it a follow-up question: How might an AI attempt to destroy humanity, i...