The Technological Singularity by Murray Shanahan

Author:Murray Shanahan [Shanahan, Murray]
Language: eng
Format: epub, pdf
ISBN: 9780262527804
Publisher: The MIT Press
Published: 2015-08-25T16:00:00+00:00

6

The Impact of AI

6.1 The Politics and Economics of Human-Level AI

We have heard a number of arguments for the feasibility of human-level artificial intelligence, either via the brain-inspired route or through engineering from scratch. We saw that, once human-level AI is achieved, the genie could be out of the bottle. The transition from human-level AI to superintelligence seems inevitable, and could be very rapid. If there is an intelligence explosion, thanks to recursive self-improvement, then the resulting system or systems are likely to be very powerful. How they behave, whether they will be friendly or hostile, whether they will be predictable or inscrutable, whether conscious, capable of empathy or suffering, all depend on their underlying architecture and organization and the reward function they implicitly or explicitly implement.

It’s hard to gauge which, if any, of the various kinds of AI that we can envisage today will actually arise. Nevertheless, we can try to think through some of the possible consequences for human society if machine superintelligence, in one form or another, becomes a reality. First, though, let’s examine some of the economic, social, and political forces that might drive or arrest its development. Why would anyone want to create human-level artificial general intelligence in the first place? The most obvious motive is economic, and a primary focus for growth is automation. Of course, increasing automation has been the trend in industry since the 18th century. But many occupations that have traditionally been immune from this trend will be amenable to automation if artificial general intelligence is developed.

The occupations in question are those that are AI-complete. A problem is said to be AI-complete if achieving human-level AI is a prerequisite for building a computer that can solve it. Passing the Turing Test (properly) is an AI-complete problem, as is (professional standard) machine translation. Occupations such as lawyer, company executive, market researcher, scientist, programmer, psychiatrist, and many more, all appear to be AI-complete. To do them competently requires a commonsense understanding of the physical world and of human affairs, as well as a degree of creativity. But if human-level AI is achieved, then it will become feasible for machines to carry out such jobs, and to do so more cheaply and more effectively than humans (as long as they can be treated as slaves with moral impunity). So there will be a powerful economic incentive for corporations to develop the required technology.

Automation is just one potential growth area for sophisticated, general-purpose artificial intelligence. New technologies can engender whole new domains of application and lead to entirely redefined lifestyles. Consider the impact of the Internet or the smartphone. Artificial general intelligence has at least as much potential to infuse our daily lives. The indispensable household robot is a staple of science fiction. But the reality is more likely to be an ambient artificial intelligence that can temporarily “inhabit” a number of robot-like bodies such as cars, vacuum cleaners, and lawnmowers, as well as accompanying users in wearable or

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.