The problem faced by AI developers is that machines don't know how to do all the work that we currently pay people $70 trillion per year worldwide to do. It requires that we solve hard problems in language, vision, art, robotics, and modeling human behavior. It does not require that whatever we build look or act like a human, or that it be "conscious", "sentient", or "intelligent" or have "qualia" or a "soul", however you might define these terms. It does not require building something that has human goals or emotions, although for some jobs it does require that it recognize goals and emotions in people and use this information to predict human behavior.
It is easier to build machines designed to do specific jobs than to do every possible job. It is also more efficient when the members of an organization (whether people or machines) specialize. Therefore, most of the work being done is in solving specialized tasks, what we sometimes call "narrow AI". Some people believe that the problem can be attacked by solving the general problem "AGI", which could be trained to perform any job that a human could do. So far, investors don't seem to agree. I believe that AGI will happen, but not by the effort of any individual or small group. Rather it will take the form of many specialists developed by global competition between self-interested people, and a communication network that allows them to share information. Because information can be shared much faster than humans can communicate, the network of specialists will look like a single entity with godlike intelligence. -- -- Matt Mahoney, mattmahone...@gmail.com ------------------------------------------- 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
