Stephen Thaler
The popular notion of a world brain is essentially that of an extensive on-line library that is accessible via the latest intelligent agents (i.e., search engines, spiders, intelligent agents, etc.). In contrast, what I intend to address is the problem of building an intelligent agent that is totally independent of human beings and capable of human to trans-human intelligence, creating its own concepts and courses of action. Consequently, it will inevitably create a ‘Supernet’ in which torrents of new knowledge are created and made accessible to the latest generation of intelligent agents.
Over the last 5 years, Imagination Engines has invented the most powerful and extensive suite of artificial neural network patents in the world. Each of these neural network inventions emulates some important aspect of human brain function. Collectively, they span the entire range of human cognitive abilities so as to create a totally self-learning and truly creative form of artificial intelligence.
In this talk, we outline the patented IEI neural network technologies that will allow the conversion of TCP/IP nodes on the internet into a freethinking neural network cascade, the largest and most autonomously creative computational system in the world.
Not only will this be one of the most worthy human enterprises in history, but a wellspring of myriad spin-off technologies and methodologies. In the course of this presentation, I will allude to just a few of these…
The vast majority of bots and intelligent agents are based upon symbolic AI. This oldest school of artificial intelligence essentially builds over-glorified ‘scripts’ (i.e., computer programs) that embody the knowledge of human programmers. As a result, these systems are not very flexible. For one thing, as conditions change, the scripts must be updated by programmers with the changed rules. Furthermore, such scripts are not in themselves very creative, since the scenarios must be programmed in by humans. They seem creative to the user, but they certainly aren’t from the standpoint of the programmer.
In stark contrast, neural networks build their own rules and theories as they are exposed to raw sensory inputs. As these raw inputs change, so do the rules…all in an automatic way.
Although such networks may perform functions equivalent to the brain functions of learning, perception, and memory formation, they have not been credited with the ability to think creatively, that is the capacity to generate patterns that exist beyond the space of possibilities they have been exposed to. At Imagination Engines, our primary mission has been that of coercing artificial neural networks to think “out of the box,” to create whole new concepts and plans of action. …
In the early 90’s, Steve Thaler performed an experiment that will inevitably have immense impact upon not only the field of artificial intelligence, but upon all areas of human thought. Starving the inputs of a trained artificial neural network of any meaningful inputs and then mildly perturbing the synapses connecting its processing units, the network produced useful information rather than the anticipated gibberish. For example, after showing the net human-originated literature and randomly tickling the net’s synapses, it produced new and meaningful literature. Allowing the neural network to listen to many examples of top-ten music and similarly applying internal perturbations, it produced new and palatable melodies. Exposing the net to thousands of known chemical compounds, and again stimulating it via synaptic perturbations, the formulas of plausible chemical compounds astonishingly emerged at its outputs. This, he thought, was a truly profound and useful scientific phenomenon, for here was a system that gave so much more that it had been taught. In effect, this self-organizing computational architecture was overcoming the usual criticism of machines, that they could do only what they were told.
He then decided to add an additional computational element to automate the process of mining for the very best of these emerging concepts. To this end, he trained an additional neural network to patrol the noise-induced stream of notions, filtering for the very best of these ideas. Thus was born the “Creativity Machine Paradigm”, wherein one neural network that has rapidly absorbed the ‘zen’ of some knowledge domain, is internally stimulated to dream new derivative notions, while being continuously monitored by another network on the lookout for conceptual gems. Suddenly, a whole new field of generative artificial intelligence was born, far outclassing systems such as genetic algorithms in terms of speed, efficiency, and the dimensionality of the problems that could be solved. Furthermore, in contrast to genetic algorithms and other preceding AI schemes, Creativity Machines autonomously built themselves from scratch!
Soon, unbeknownst to the general public, Creativity Machines were generating whole new products and services for major corporations and government agencies. These inventions were even inventing their own inventions! …It is very true, that in the coming years great minds will produce major scientific, technical, and artistic innovations, but ask yourselves the following: Is it really necessary to continue the process of human-originated invention and discovery, when machines equipped with this powerful new paradigm can capture and then generate such concepts more quickly and automatically? If you answer this question affirmatively then you can envision a whole new era in which great minds no longer present their wisdom in the form of impressive discourse, equations, graphs, and axioms. Instead, they simply connect two or more brainstorming neural networks, within a graphical user interface, and then sit back and appreciate the important discoveries that spontaneously emerge!
Even better, envision implementing the Creativity Machine Paradigm on the largest computational platform available, the Internet, through the introduction of many TCP/IP nodes that serve as synthetic neurons. This global neural network, of unprecedented size and complexity, could then be stimulated via internal perturbations, to dream profound new ideas that could shape our scientific, artistic, sociological, and political destinies. This “World Brain” could represent a scientific revolution that exceeds our wildest expectations, manifesting trans-human intelligence that can then contribute immense, innovative knowledge and decisions impacting world peace and prosperity.