in many subfields of AI, the stuff that's locked up in Google proprietary information is light years beyond what's available in academia
What is your evidence for this? (Sorry if it's somewhere in the reddit thread, I didn't read too far down.)
I have heard this claimed by multiple sources but looking at the webpages of most google research scientists indicates that they aren't even working on new theory so much as applying what's already out there, so I'm curious what's causing our beliefs to diverge so much.
I don't have any evidence beyond Jonathan Tang's say-so. And the fact that Peter Norvig works at Google. There may be something useful in Reddit's video interview with him, but it's been ages since I watched it, so I don't know.
I have been having some difficulty deciding what to do with my life. I'm not really ashamed or surprised, because the problem seems extremely difficult and worth getting right. I still don't really know. Here are some of my options, though I wouldn't be surprised if what I end up doing is not on the list. I thought I would share to elicit advice, to give some context to some of my recent remarks, and maybe to provide comparison for people in similar situations.
1. Research
I could do research in a university, or in a private research lab. Many fields have at least a few questions that seem important and interesting.
A. Theoretical Computer Science. I could work on collaborative learning, recommendation and reputation systems, distributed protocols, or anything else that might assist collaboration in the future.
B. AI / Computational Cognitive Neuroscience. I could work on algorithms for inference and planning, to increase the probability that we develop comprehensible or human-like AI before something horrible happens (like developing incomprehensible strong AI).
C. Neuroscience. I know almost nothing about this field, but technology for measuring and interfacing with the brain appears to be important and developing rapidly. I don't know how hard it would be to start working in the field, but I suspect that the connection to computer science is strong enough at MIT that it wouldn't be impossible.
2. Start a company
This seems way harder, but I am sufficiently arrogant and risk-neutral that I consider it a reasonable option. In particular, this is what I would do if I decided that making money as efficiently as possible and giving it away
A. Tech company. If I had to guess based on the currently available evidence, I would guess that this is the way to maximize my expected earnings.
B. Online Education. I would like to take a shot at designing materials for online education of smart, significantly underserved, high school students.
3. Cooperate with Other Rationalists
A. Work for a rational charity. Self-explanatory. Probably worse than earning a lot of money and giving away.
B. Start a rational charity. Probably worse than supporting an existing charity.
C. Work for a rational start-up. Can't really arrange this one; but optimistically it could happen if you were prepared (someone else does 2).
4. Be a Hobbyist
I could also simply try and earn a living as quickly as possible (rather than making as much excess as possible, or having a more structured way of doing good) and do work on the side. I don't think this is a good idea.