Quote:
|
I think peter12's real issue is against those who think AI will be able to take over for humans in systems where decisions need to be made based on rational understanding, intuition, and maybe common sense, over a more mechanistic approach. I agree with this stance - without understanding the limitations of AI, it can be misused and result in very unforeseen consequences. However, I think the recent increase of interest in AI is a good thing, even if that interest is misplaced. AI was considered a near-dead field when I was in university, but recent advancements have vastly improved the quality and "learning efficiency" of neural networks leading to, among other things, the defeat of a Go master.
|
You seem like someone with a pretty strong background on the science side. I will defer to you on all those questions as my credentials there are exactly nil, and my potential to gain real understanding of the algorithms, and all the high-level mathematics involved in machine learning is also zero.
My real view, as you precisely identified, is a concern as to the personal teleos or purpose of human beings in light of the extensive promises and threats that artificial intelligence could have for our species. We are remarkable creatures, both flawed and divine, and it is simplistic to believe that machines - or as some in this thread have said, human-created GODS - will solve the cipher of our existence. If they are able to do so, and not tell us, then what is the point?
I am not really sure if there is anything further to discover. Certainly, the utopian promises of Ray Kurzweil et al. are without foundation. We are not anywhere near decoding our neural code, understanding human consciousness, or even scratching the surface of our own nature. It seems highly implausible that we will discover or create anything approaching artificial general intelligence without those key pieces of information, and if we do, the dangers, both imagined and real, are worth contemplating.
But beyond that, I really think we need to be honest about what machine learning can actually do. As I said, I do not understand the algorithm side, and I would really find your insight helpful, but it seems that we can create increasingly sophisticated routine-based machines. You are right. Defeating a Go master is significant, and maybe the long-term predictive abilities of that machine will be able to have some use outside of a Go championship game.
To me, right now, it is akin to IQ testing. We know that you can practice the heck out of IQ problems, but when you take an actual test comprised of secret answers, you fare slightly or no better than someone who didn't practice at all. We aren't creating machines that are able to spread their intelligence across multiple tasks, and this, of course, is the essence of real artificial intelligence.
The question remains, does knowledge about intelligence fall into the realm of diminishing returns? Is it like so many things where the more we know, the less we are able to know.