The (IM)-Possibility of AI (Humans)

by xeno6696 @, Sonoran Desert, Sunday, October 17, 2010, 21:46 (5128 days ago)

Having spent some time in my hiatus to collect and condense some more of my thinking, I spent some more time getting myself acquainted with AI research, but more specifically, the hardware limitations that underpin AI. -To more bluntly answer dhw's earlier question, "Do I think that AI is possible?" The answer is: Not with current technology. -All of the current AI applications that I have seen are all special-cases; a machine can be trained to solve a particular problem, I think that the binary nature of machines prohibits them--even considering things like "fuzzy logic," (a technique where machines make decisions based on probabilities) current machines can only provide binary answers to questions. One of the largest arguments I came across for this is Martin's direct assault: No equation has ever been written that can solve all problems. This seems minor, but even AI, at its core, solves problems using some kind of predetermined mathematical formula. So from one standpoint, a "general AI" would in some way need to be able to take in the entire world via some singular formula "for everything." -Where I lift this embargo on true AI is when quantum computing gets involved. Machines will no longer be limited to binary (Yes/No) distinctions. A "formula" as it were, need not exist at all while nature's probabilistic dice tumble in its head. Quantum computing will no longer rely exclusively on complex mathematical models--a quantum computer is working with the actual "stuff" of the universe, and not a binary abstraction of the universe that we have with current digital technologies. -The over all difficulty for general AI lies in problem-recognition; humans are notoriously bad at this at a fundamental level. How do you pick the right solution for the right problem, and how to you pick the right solution for an unknown problem, or in other words, how can you know that a "good enough" solution exists? -I don't think this means we should stop trying to build human intelligence with current technology, however I will say that at a very basic level, there is very large problems in building a generalized AI much less a conscious one.

--
\"Why is it, Master, that ascetics fight with ascetics?\"

\"It is, brahmin, because of attachment to views, adherence to views, fixation on views, addiction to views, obsession with views, holding firmly to views that ascetics fight with ascetics.\"

RSS Feed of thread
powered by my little forum