CH 51: A.I. Inflection Point
One of the most intriguing and thoroughly overlooked organizations could end up being the Singularity Institute. Their primary goal, esoteric as it might sound, seems to distill down to “seeding good A.I.”. You are probably asking what on earth is that suppose to mean. Well, A.I. of course refers to Artificial Intelligence, which we know can be a rather broad term. After all, there is A.I. in vacuum cleaners, cars, planes, and toys, to name but a few. In these commonplace examples, the A.I. has a very narrow domain of abilities, but truly excels at those abilities. Where A.I. is weak is in a parallel processing context, where it is about being able to tie together very disparate concepts effectively, and where humans really excel.
However, it is a reasonable assumption that over time, whether it is minutes from now or decades from now, the software underlying certain Artificial Intelligence will have in its code the recursive ability to improve on its own code. This means that future A.I. will actually be able to rapidly improve on itself. This remarkable likelihood means in certain theories that A.I. could very quickly develop the ability to more efficiently and effectively identify and solve problems than humans do. If such an event does indeed happen, it would mark a singularity event since the brain of such A.I. could be more effective than humans and since we can't predict outside of our own capabilities, we would have no idea what such a future would look like, hence the term singularity.
The purpose of the Singularity Institute is to do all it can to lay the foundation and conditions for the best possible good A.I. to evolve and therein not have any desire to destroy mankind. While this idea seems more science fiction than fact, this organization is firmly rooted in mathematics and science. It is just that such a topic is so intellectually arcane that most people have more pressing things to consider, rather ironic considering the stakes if the good people at the Singularity Institute are indeed correct. The key to their success is mathematical breadth and depth and the ability to convert knowledge into practical good A.I. developments. Ultimately, the language of math will best describe the A.I. that this organization seeks to build. Speaking of current narrow-domain A.I., I.B.M. has been hard at work readying its supercomputer Watson for the challenge of playing the classic question and answering game Jeopardy! against human competition. For more information on the future, see the Singularity University.
