If Deep Learning is 'It' then why are search engines in decline? With all the computing power that the industry has at hand search engines should be getting a little smarter every day. Therefore I have come to the conclusion that the efficacy of Deep Learning is at least partly and maybe mostly hype. I do acknowledge that there have been major advances in AI it is just that they do not add up to the kind of general intelligence that people are capable of.
It is true that programs like Watson and Deep Learning programs are capable of being applied to a wide variety of problems so to say that they do not constitute general intelligence just does not make sense. Their applications may seem narrow but there is no question that they could be refitted (so to speak) to deal with a greater variety of problems So I reject the statement that they are just forms of 'narrow AI' because that kind of criticism no longer reveals what is missing in the contemporary paradigms. We have to form a better criticism. My criticism is that they lack the power of insightful conceptual integration. If Deep Learning is not 'It' then why not? I believe the answer is probably simple. Deep Learning is an extension of hybrid machine learning and there is not anything new there (at least of the sort of thing that we are looking for.) From this point of view I would guess that Watson did use deep learning (with small d and small l if you will). I also believe that programmers could take something like a chess program and apply many of the same principles to a variety of different kinds of situations. Then work it on a massive corpus of situations and the responses that it produced could be sampled and scored. (I am not advocating this as an AGI program I am simply saying that this methodology could be implemented in Deep Learning.) So my conclusion is that Deep Learning has not presented us with some new paradigm (of great importance to our interests). It is essentially an extension of systems that have been a part of machine learning for a number of years. So the attempt to 'prove' that I am wrong about Deep Learning and 'narrow AI' is not very interesting without a very well structured argument. It is not that Watson did not use Deep Learning and that Deep Learning is narrow AI; the problem, in my opinion is that none of these methods detail how conceptual integration might take place in an AI program. Deep Learning may be new but it is not imo something that has produced a program that is distinctively more intelligent. Analogically, the argument that Deep Learning is narrow AI is not a criticism that sheds light on what is missing in these AI methods. It is an old criticism that is true in some cases but falls short in most cases. ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
