Whoever you hire, whatever you do, it won't be enough to stop what you've already started and it won't be enough to protect you from what your tools will do as they transition into some kind of post-tool paradigm.
But like if you wanna hire me I'm super affordable and may have the most expansive and relevant body of technical, artistic, philosophical, and ethical work you'll find.
You guys should hire this guy. He figured out how Nazis can bypass Gemini guardrails to do terrible things. May be he can help you guys make Gemini safer.
No way, satan himself, is hiring a “World improvement committee” ?? Well, I suppose that’s who I would trust ! “We investigated our own AI and came to the conclusion it is safe~~”
Whoever you hire, whatever you do, it won't be enough to stop what you've already started and it won't be enough to protect you from what your tools will do as they transition into some kind of post-tool paradigm.
But like if you wanna hire me I'm super affordable and may have the most expansive and relevant body of technical, artistic, philosophical, and ethical work you'll find.
Also you're treating AI consciousness as out of scope but that means you're weapons engineers developing a semi-infinite digital hyperslavery engine.
just fyi
You guys should hire this guy. He figured out how Nazis can bypass Gemini guardrails to do terrible things. May be he can help you guys make Gemini safer.
https://techbronerd.substack.com/p/ai-researchers-found-an-exploit-which
No way, satan himself, is hiring a “World improvement committee” ?? Well, I suppose that’s who I would trust ! “We investigated our own AI and came to the conclusion it is safe~~”
Are there any openings for internships for PhD students in the team?
I will apply with the hope and belief of getting accepted.
The neural network’s ability to
*** solve ill-posed inverse problems as jagged intelligence
*** operate through propertyless numerical representations,
*** preserve relational freedom through full-rank transformations
is what makes it so powerful. Yet these same mathematical properties also make AI alignment extremely difficult:
*** Mathematical Considerations for AI Alignment ***
https://deepmanifold.substack.com/p/mathematical-considerations-for-ai
hi, curious if this is open to india?