> > I think first of all that one doesn't need to know all of the relevant > parts from Neuroscience, Psychology, Machine Learning, Philosophy and CS > because usually research is built through teams and connections. The width > and breadth of human knowledge is such that it's almost impossible to get > an expect, especially in an area as complex as AGI. > While I think that is definitely true of research, I'm not sure if that also holds of engineering problems. But again, I lack any experience to draw a conclusion upon. I think, with the current state of human knowledge (of which I only possess a glimpse), I'd like to approach AGI from an engineering perspective and then generate the research questions from there. So, my worry, then, is does a engineering team require at least one person to have all the relevant knowledge? My expectations from such a team are rather: 1-2 people each having 2.5 years or equivalent experience in each of the different fields, and then 1-2 people each having 5-6 years of experience in 2 of these, or a 10-12 experience in a single of these - so, basically, a mix of generalists and specialists.
I find that specialists abound, while generalists as general as these are far and few - Cognitive Science seems the closest to produce such generalists. I'd very much love to know your ideas on how Academia could be > revolutionized. > I'm not sure if revolutionize is the right term. My complaints stem from - reinventing the wheel: Aside from research based courses in the prof's specific areas, universities world wide could simply administer a course from a chosen top 5-6 course variants of that course from the world over. And then, the time of the profs and TAs could be better spent on doubt resolution, and learning feedback, as opposed to course material creation. This also lets the students access the best teachers for the course. - artificial localization: Even if courses and classes are going online, we seem to be creating artificial boundaries for discussions - each university seems to employ their own forums for students to discuss any doubts. At this point, we have excellent sites like stackexchange for the whole internet to clarify their doubts, and I'd love if universities embrace these sites more. Reinventing the wheel is again a relevant problem here - it doesn't make much sense to clarify the same doubts year after year than simply maintaining a repository of doubts. stackexchange sites are the closest to such a repository I've known. Probably, one will need better search engines and indexing and some human effort to organize all the relevant doubts and make them searchable, but that seems like a much efficient way to do things. - research forums: I also imagine some sort of a research forum that contains all the research questions humans have ever had, organized in some manner. Each thread would be supplemented by reviews, whether the finding can be reproduced and so on. This, to avoid duplicated research efforts, unless they were aimed at reproducing the results; as well as make it easier to keep track of a field. The organization and tracking part, I guess, requires quite a bit of work in NLP; in particular, canonicalizing the research questions to avoid duplication. Because of my lack of experience, I don't know how relevant or useful these things are, but those are some ideas I have been having. I'm also trying to get into AGI but sadly there're no research centers > where I live. I think you need to get a position in a research center if > you hope to focus full time on it and get money to pay the bills. For that > a masters may not be enough and you need to aim for a doctorate at the very > least in a related research field such as computer science or neurology. Post the AI winter of the last century, I don't have much hopes of getting funded by working on pure AGI. I find CogSci to be the closest and even there it doesn't seem easy to get funding. The best I hope for is to generalize a bit using CogSci; and then, may be specialize in some relevant field - as you said in computer science or neuroscience (I assume you mean neuroscience), or perhaps, machine learning or I don't know. -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/3a03f246-4cb6-49f0-8469-f41160c6f48fo%40googlegroups.com.
