J. Storrs Hall, president of the Foresight Institute, takes the position in this debate that the rise of artificial intelligence levels will create a utopia for humanity. Hugo de Garis, Xiamen University, China, takes the opposite position, namely that the rise of godlike massively intelligent machines will be catastrophic for humanity, leading to the worst, most passionate war humanity has ever known, using late 21st century weapons, killing billions of people.
Part one: Hugo de Garis
Part Two: J. Storrs Hall
Future of AGI Debate - Rebuttal from Jeriaska on Vimeo.
Share this article with:
Fascinating Debate ... my hypothesis
This is very fascinating. I would like to point out that artificial intelligence, in and of itself, is neutral on the morality scale. Obviously it can be used for making war and it can be used for bringing prosperity to mankind, and de facto they will be used for both too. I just question how much for each purpose? I certainly hope that artificial intelligence will be used mostly for the purposes of good, for curing world hunger and bringing potable drinking water to the masses. Certainly artificial intelligence will usher in a new era of ever improving ecconomic prosperity and will enable humanity to explore and excavate the regions of space for purposes of ecconomic and social gain as well. I do, however, forsee a future where human laws and human inventions regarding government are obselete, much as religion has become obselete. The future is in business and technology, and in order for that to work to our advantage there must always be humanity in between the two, moderating.
It is fighting within the order of beaurocracies as I see it.....religion, government, business, and technology....whereby each successor seeks to eliminate the previous and each previous seeks to control and regulate the successor. I would, indeed place humanity and education, and also artificial intelligence at a place in between the business and the technology spectrum on the scale of beaurocracies. So long as, say, a successor doesn't control a prior (government controlled religion or business controlled government or technology controlled business but human controlled business), then we are on the right track.
AI is less of a threat than people are.
Our largest sources of danger today are people.
The reason people are such a problem is that we have evolved to have aggressive tendencies due to the need to compete for resources, to fight predators etc. This combined with the fact that our intelligence is limited and our reasoning ability is generally poor.
Artilects, to use De Garis's term, will generally only be a threat to us if they are deliberately designed to be a threat. Unlike humans, an artilect would not be aggressive or have self interest unless it was designed to be that way.
There are many ways that we might deliberately or inadvertently build an aggressive artilect, e.g. by replicating the human brain or by using evolutionary processes where self preservation and aggression might be advantageous.
But there are also many potential ways of designing artilects that are not likely to produce artilects that are a threat to us. E.g. via evolutionary processes where fitness equates to specific traits that we want the artilect to have, e.g. the ability to understand mathematics, or the ability to understand natural language and to be able to store knowledge and answer questions. There is no reason that an artilect can't be more intelligent than a human being and understand humans and our emotions and yet have no emotions or self interest of its own.
The main reason that people have a fear of artilects is that they think of artilects as artificial humans and having all of the same personality traits that humans do. I think it is much more likely that most of the artilects that we create will not have human like "personalities" and they will not generally have the same goals and "desires" that humans have. We must be very careful in choosing what desires and goals we give our artilects. I think it is very dangerous to attempt to build artilects by directly mimicking the "design" of human minds.
Human motivations are ultimately determined by our evolutionary origins, and that is the cause of our biggest problems today. There is a legitimate danger that cyborgs might pose threats to humanity, but I think it is possible to have a wonderful future that transcends our current limitations as evolved animals, but we need to be careful about how to bridge the gap between humanity with all its failings and an improved humanity with improved intelligence, reasoning and social responsibility.
The Possibilities of Artilects
The way that I see it is that any life form that self replicates will develop a need for obtaining and eventually controlling energy and resources. Wherever there becomes a competition for limited resources, life can either be forced to ration and share the resources or it can fight for the control of those resources.
The first route, rationing, leads to a lower initial payoff while the longterm has the potential for survival through cooperation.
The second route, fighting, leads to a higher initial payoff assuming the fight was won, but the potential longterm consequences could be in eliminating key characteristics of the population that may be necessary for survival.
In essense, fighting rids diversity but allows for a thinner segment of the "herd" to flourish, while cooperation tends to integrate diversity which may or may not be providing adequate resources for the "fittest" within the herd. Of course, this does not look into a quantitative analysis for cooperation, it is pheasable for rationing to be done in proportions that are suited for each individual part of the heard via a hierarchical structure, but eventually there will be aspects of the herd that will want to break off and compete against or within that herd in order to produce or obtain their own wealth and resources.