Dima raised a valid concern: “in what ways is a MOOC an improvement over a well-annotated reading list like the one of CHAI?”
For reference: http://humancompatible.ai/bibliography
I realize I have a strong (sunk cost, status) incentive to rationalize that RAISE is actually useful. I tried to account for that.
Yet I did come up with a few reasons why RAISE is a worthwile improvement over a reading list:
– impartiality about which approach/institute works, taking them all into account
– social motivation structures which are a necessity for at least some high-potential individuals
– knowing which people are talented, so we can approach them. Not everyone’s confidence matches their aptitude, so some high-potential people would never dare step in by themselves.
– related: being invited counters imposter syndrome, which can be a serious distraction
– a learner’s hub can help learners in more ways than just a MOOC. A charity that specializes in running such a hub will probably be more effective than a charity that specializes in research, and does some learning aid on the side.
– aggregating everyone at one place makes coordination better and everyone will have at least an introductory understanding of everyone’s lines of attack on AIS
The crux of some of these points is that it takes a very specific entrepreneurial character to learn everything yourself. I suspect that one of the reasons no one has already done what we’re doing is that those who start a non-academic research institution are suffering of typical mind fallacy. They’re super bold trailblazers that don’t fully realize the extent of social proof most people rely on.
You could of course argue that this is a good filtering effect, since we actually want people that can come up with groundbreaking ideas. I have a suspicion the tendencies to make your own path and come up with your own ideas are strongly correlated.
But even if we only want that kind of people, even they would benefit from being served with support.
But then a well-defined path with social support doesn’t really require that we rehash all the content. Why don’t we just start with a reading list instead of filming lectures? Lean startup style?
Opinions are very welcome!