Search for a non-human form of intelligence – From “SETI” to “Super-intelligence”
In the 80s, the science fiction novel “Contact” fascinated millions, and when in 1997the movie based on it was released, it became a blockbuster. The theme of Carl Sagan’s fictional work is mankind’s first contact with an alien intelligence: After humans receive via electromagnetic signals a message by an extraterrestrial intelligence which entails instructions for building a machine enabling space-time traveling, the protagonist experienceswhat it means that we humans are only an insignificant and not even highly developed intelligence in a cosmos full of such. The story ends with the discovery that in the number pi is encrypted a message that points to a higher power, so ultimately a God. May many SETI (“Search for extraterrestrial intelligence”)-fans have been disappointed by the rather cheesy religious finish of the story, Carl Sagan nevertheless illustrates with beautiful prosethat a contact with an extraterrestrial from of intelligence would likely represent the most significant event in human history. A real “game changer”, we could say.
To date, the book and film have lost none of their fascination. The imagination of science fiction writers (Carl Say was also a serious astronomer and astrophysicist) has always been an indication of how technological possibilities at the time inspired humans. Taking the novels and films of the more recent past to that scale, we find a subject related to SETI moving our minds and souls today: the development of a higher artificial intelligence (“AI”).
Already back in the 1950s, computer scientist and other researchers were driven by almost limitless expectations regarding the capabilities of computers and the possibilities of AI. They believed that within ten years a computer would beat a World Chess Champion and prove an important mathematical theorem (the former did not happen until1997; however, the latter expectation was fulfilled already in 1976, when the so called four color theorem was proven with the help of a computer). In 1993 the mathematician and computer scientist VernorVingemade the prediction that “within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended.”So according to Vingethere are eight years left for us, plus maybe a little more. Forecasts like this, similar to the one made by Ray Kurzweil on an imminent “technological singularity” quicklyclaimsignificant public attention (and find lucrative positions for those who publish them: Ray Kurzweil for example became head of technical development at Googlein 2012).
Such an intelligence which would significantly exceeds human cognitive performance in virtually every respect (not just playing chess) we call a “super-intelligence”. Based on its own skills it could developscience and technology even faster, accelerate technical progress even more rapidly,and in turn create even more intelligent systems. There would thus be a feedback loop and likely such a rapid (exponential) technological progress that humans with their limited cognitive meanswould no longer be able to follow.
The logic underlying the technological singularity-thesis is rather simple: According to “Moore’s Law” the power of computers doubles about every 18 months. It is thusonly a matter of time until their processing power outstrips that of the human brain, the computing power of which can be estimated to be up to 20 petaflops (1 petaflop corresponds to 1015 – that’s a 1 followed by 15 zeros – point operations per second). In May 2008, the company IBM was the first that broke through the petaflop barrier with its computer “Roadrunner”. Since June 2013, the Chinese computer “Tianhe-2” holds the record as the fastest machine in the worldwith 33.9 petaflops. The goal of the companies involved in this technological arms race is to build anExaflop computer (1018 operations per second) by the end of the decade. Its processing power would be approximately 50 to 100 times higher than that our brains.
But powerfulhardware alone does not suffice for a true super-intelligence. Each computer requires a software. For a technological singularity in the form of a super-intelligence we would need to create a KI, which thinks creatively about problems and is able to solve them autonomously.We may need a different computational technology to achieve this, possibly one that functions closer to our brain (massive parallel processing of information) instead of along the lines of the classical “von-Neumann” architecture (sequential information processing: single instruction, single data). Already in the 1960s, the American computer science pioneer Joseph Licklider’soutlined avision of a massive network of the human brain and computers, as well as a global computer networks (which comes surprisingly close to today’s Internet). The term “cognitive computing” today describes a possible future generation of computers with human abilities such as adaptive learning based on feedback loops, flexible reactions to external sensory inputs, and social interaction with other systems, all things that are very difficult for today’s computers. Computer engineers expect the next big technological evolution to occur in this area. Its applications would be diverse. Dealing with confusingly many and complexly interrelated data, be it in the medial field with the interaction of drugs with our metabolism,in analyzing the huge amount of information at the LHC at CERN, for planning the optimal transport system in urban societies, or for an understanding of the dynamics of global capital markets, demands increasingly powerful computational means, where current computer technology constantly reaches its limits.
Of course, we do not know the path of future technologies. In addition to “true AI” there are several other ways towards a super-intelligence, such as genetic manipulation of those hereditary units responsible for our intelligence, powerful interfaces between the brain and computers or even a full integration of the neural architecture of our brains with computers, or last but not least the spontaneous development of a collective network intelligence (keyword “intelligent internet”). But the first technological steps towards such developments have already been made. Based on the current possibilities of computer-, bio- and neurological technologies and their developmental dynamics it is by no means absurd to assume that the creation of a super-intelligence will occur in the course of this century. We should rather count on it. Such would potentially be unimaginably powerful and come with enormous monitoring and control problems for us humans, compared to which our current economic, environmental, and social problems would be almost negligible.
Reflecting on such a development we quickly encounter questions that we cannot yet answer today. Would a super-intelligence possess a consciousness of itself (we would here find interesting connections to an age old philosophical enigma)? Would it be subject to suffering? Would it have an ethical and moral framework? The answers to these and many other questions are critical to assess the impact a super-intelligence would have on human civilization. But even without those answers, we can easily say that for humanity the creation of a super-intelligence would be at least as impactful (and threatening), likely even more powerful (and scarier)of a “game changer” as a first contact with an extraterrestrial intelligence. Unlike such we can meanwhile outline a very realistic path towards a super-intelligence. We would be well advised to deal with the related issues and possibly launch a program, analogous to the various SETI projects, to detect any its germsearly enough. Because once it is there, its momentum for further growth can develop astonishingly fast.