"AI can do a lot for medical discovery, but there are also legitimate concerns that AI is outperforming PhD-level virologists, & that could abet a bioweapons risk that is not hypothetical."
- Johns Hopkins professor & adviser to Anthropic, Ben Buchanan
Some users stress the dual-use risks of AI outperforming virologists as a reason to demand stronger safety governance, while others reply with accusations of malice toward figures like Dario and insults dismissing concerns as childish.
@rohanpaul_ai are any of these people doing the first thing to harden society against a pandemic? this information is ALREADY out there, and hostile governments already know about it this cat is out of the bag RIGHT NOW, AI or no
@rohanpaul_ai AI's potential seems matched by its risks
"AI can do a lot for medical discovery, but there are also legitimate concerns that AI is outperforming PhD-level virologists, & that could abet a bioweapons risk that is not hypothetical."
- Johns Hopkins professor & adviser to Anthropic, Ben Buchanan
@rohanpaul_ai This highlights the dual-use nature of advanced AI. Capability gains in science also increase the importance of strong safety and governance frameworks.
@rohanpaul_ai Dario you are even more of an evil genius that I ever imagined. First, you scoop up all the 20,000+ 20 year old zero day viruses that have survived human discovery, now you employ the best and brightest virologists to learn how to attack nation states and the world is your oyster
@rohanpaul_ai Fund me to work on this "....not hypothetical ...." risk. But why go to the expense? We have COVID, Flu (including bird flu), measles, chickenpox, and others. Just a lapse in sanitation, and you can add cholera and typhoid.
@rohanpaul_ai An AI that outperforms PhD virologists is not hypothetical risk — it is the exact same capability curve that made every other powerful dual-use technology dangerous before governance caught up. The honest answer is nobody knows where the line is until someone crosses it.
@rohanpaul_ai This question demonstrates the childishness of this TV entertainment personality.
Apparently she believes that pieces of paper with words written on them, she describes it as a "treaty", have any meaning or authority at all.
Experience teaches these papers are meaningless.
@rohanpaul_ai Artificial Incompetence can also do this.
@rohanpaul_ai Doesn´t this same risk exist for every PhD trained virologist?
"AI can do a lot for medical discovery, but there are also legitimate concerns that AI is outperforming PhD-level virologists, & that could abet a bioweapons risk that is not hypothetical."
- Johns Hopkins professor & adviser to Anthropic, Ben Buchanan