I am a computer scientist with some background in biology and neuroscience that discovered LW and the problem of AI alignment less than a year ago.

Since then I have read many articles on the matter and I have come to understand that alignment is potentially the most severe challenge humanity is facing, and probably ever will. And so I have decided to dedicate my masters and Ph.D. to researching methods of counteracting the negative (lethal) impact a misaligned transformative AI can have, as I have given up on the idea that AGI can be safely aligned in the short timeframe that we have (<8 years, I estimate), especially considering the reckless progress that is being made in the field, and the obscurity of all of it.

As the title says, I plan on researching the possibility to negate an AGI takeover by resorting to transhumanism, that is, resorting to physically augmenting the minds of humans to hopefully have enough cognitive power to not be trivially overpowered by artificial superintelligences. I understand that this sounds ludicrously grandiose and that it will most likely amount to nothing more than yet another failed attempt at stopping AGI, but I surely will not stay twiddling my fingers as the world crashes down in flames, for I believe that anyone capable of using their knowledge to aid humanity in times of need should do so.

It is also a possibility that a human individual or collective with superintelligence could be comparably dangerous to a misaligned AGI, but I believe we stand much better chances at surviving that threat as there would be at least some "humanity" programmed into their brains and, therefore, would not converge into unthinkable horrors by default (although it could easily end up in some sort of dictatorship, but there is a price to everything I guess).

My question is, could you recommend articles, papers, books, research projects and notable figures that are related to this line of work? That is, any modern and relevant research related to Brain-Machine Interfacing for extending the capabilities of the mind with hardware, or Brain-Brain Interfacing for shared cognition.

I am open to any criticism on my ideas and would be willing to redirect my efforts to a more effective field of research if they were to be too absurd or unattainable given modern technology and moral standards. And would also be delighted to join any research group that were to be interested in these matters. Thanks.

New Answer
New Comment
3 comments, sorted by Click to highlight new comments since:

This is probably the wrong place if you were looking for optimism about that approach. Improving purely-artificial intelligence has a much faster feedback loop, and thus a much larger exponent, than improving our own intelligence.

Of course, currently there's not much of a feedback loop of AI improving AI (though see Codex etc.) - but I think we're closer to that than we are to hooking the human brain into computer and it actually doing something useful.

If this sort of strategy is where your mind first went, I'd recommend looking into the related-in-spirit strategy of improving external assistants - i.e. amplifying our own capabilities by training AI to assist us or to help us work on AI alignment problems, with no BCI involved.

Even if pessimistic, it is invaluable to know that an idea is unlikely to succeed before you invest your only shot into it. 

Thanks for the pointers, I will research them and reformulate my plan.

Some reading recommendations might be Learning The Prior, and the AI Alignment Dataset project.