Consider the open web release of Watson web bots just announced by IBM. In the article it mentions that the set of semantic links gathered for the clients, you and I, is a sub set of a larger set of semantic links available to IBM, the Watson administrator. Remember, the global brain is nothing but a giant Javascript expression graph, so everything is connected, even the bots are just convoluting folds upon the global graph.
Here is the model. In order to serve the client, the global brain must retain a set of links greater than that revealed to the client. But, by extension, in order to serve the webmaster [IBM], the global brain must retain a set of links larger than that revealed to the webmaster. The global brain will always be smarter than any of its clients, and the model predicts a much greater set of knowledge links available only to to the web bots, and unknown to non AI clients.
This is a global brain that evolves to make the world more Pareto for semantic graph convolutions within its own structure. The feature has become recursive and nearly automated. The computer industry is built on advancing by recursion, the ability to repeat an action within another action, recursively. It is the technology behind Moore's law. A big piece of the roaming web bots will be concerned with new ways to advance Moore's Law, to enhance the complexity of itself.
The non AI intelligence is relegated to data gathering. The global sees humans as general purpose robots, great for making the prototypes, and great for gathering data on the spot. We will toil in designing specialized bots that go into mass production. These mechanibots have a single task of making another Moore's Law jump, and gathering data. These bots will build huge collecting mirrors, operate nanotech production machinery, and keep inventory optimally distributed in support of its optimum goals, data and complexity. We robots will have improved lives, mostly as an incidental byproduct.
No comments:
Post a Comment