março 19, 2004

SINGULARITY WATCH

(...)computer complexity has been increasing at a double exponential pace for over a century now, without exception (actually, for billions of years when we consider molecular, cellular, and organismic "computers"). This means the rate of acceleration is itself gently accelerating, while modern computer design becomes ever more human-independent, allowing us to forsee a future of emergent Autonomous Intelligence (A.I.). For the first time in human history, we have a rough quantitative sense of when our double accelerating computer and our finite and bounded human complexity curves will roughly intersect. Some project a "self-evolved" human-equivalent machine intelligence will arrive by 2030. Others by 2070.
(...)since the late 1960's, the fields of scientometrics and informetrics have been measuring the growth rate of various informational parameters in technological society. It has become clear that the "doubling rates" of scientific and general information are, like computer complexity, also progressively shortening each year (down from 15 years in the 1960's to 5-7 years or less by 2001, for several parameters). Some observers have projected that machine complexity will continue to drive these rates ever faster, until a time when the rate of change will appear effectively instantaneous to a biological human observer. Interestingly, this projected date also occurs circa 2020 to 2060, providing yet more independent—though still circumstantial—evidence that a transition of deep significance to biological organisms is rapidly headed our way.
(...)the evolutionary computational developmental paradigms of our applications, agents and robots in coming decades may be engaged in replicating the entire metazoan evolutionary developmental learning curve, taking our digital systems from insect to human level intelligence within approximately 40 years. Most importantly, if these increasingly intelligent computational systems are ever more balanced and integrated with human society, and providing increasingly useful solutions to human problems, humankind will allow this evolutionary learning process to continue unabated. As environmental inputs drive evolutionary learning, human guidance will be only minimally required in this process, perhaps primarily to ensure developmental safety.
(...)If all of this is true, even partially so, how will we create this merger, how will we enter this Symbiosis Age (an apparent successor to the Information Age) in a balanced and sustainable fashion? How will we protect both our humanity and our deep respect for the biological world? How do we ensure the stability, diversity, and spirit of world culture as it undergoes this apparently inevitable accelerating transformation? Will there be issues of safe artificial intelligence in the development process? Or will we find, as many now suspect, that technological systems are not in competition with biological ones, but will rapidly occupy entirely different niches? As our technologies begin to wake up, how do we use them today to improve the immune systems of human societies, in a world with access to ever more effective weapons of mass destruction? How can we continue to increase our scientific, personal and financial foresight in a world of accelerating change? Which sciences, technologies, businesses, and humanist practices will aid and inform us during this accelerating transition?

Sem comentários: