Редактирование: Transhumanist FAQ Live (раздел)

===Will posthumans or superintelligent machines pose a threat to humans who aren’t augmented?===

Human society is always at risk from some group deciding to view another group of humans as fit for slavery or slaughter. To counteract such tendencies, modern societies have created laws and institutions, and endowed them with powers of enforcement, that act to prevent groups of citizens from assaulting one another. The efficacy of these institutions does not depend on all citizens having equal capacities. Modern, peaceful societies have large numbers of people with diminished physical or mental capacities along with many other people who may be exceptionally physically strong or healthy or intellectually talented in various ways. Adding people with technologically enhanced capacities to this already broad distribution of ability would not necessarily rip society apart or trigger genocide or enslavement.

A common worry is that inheritable genetic modifications or other human enhancement technologies would lead to two distinct and separate species and that hostilities would inevitably develop between them. The assumptions behind this prediction should be questioned. It is a common theme in fiction because of the opportunities for dramatic conflict, but that is not the same as social, political, and economic plausibility in the real world. It seems more likely that there would be a continuum of differently modified or enhanced individuals, which would overlap with the continuum of as-yet unenhanced humans. The scenario in which “the enhanced” form a pact and then attack “the naturals” makes for exciting science fiction but is not necessarily the most plausible outcome. Even today, the segment containing the tallest 90 percent of the population could, in principle, get together and kill or enslave the shorter decile. That this does not happen suggests that a well-organized society can hold together even if it contains many possible coalitions of people sharing some attribute such that, if they unified under one banner, would make them capable of exterminating the rest.

To note that the extreme case of a war between human and posthuman persons is not the most likely scenario is not to say that there are no legitimate social concerns about the steps that may take us closer to posthumanity. Inequity, discrimination, and stigmatization – against or on behalf of modified people – could become serious issues. Transhumanists would argue that these (potential) social problems call for social remedies. (One case study of how contemporary technology can change important aspects of someone’s identify is sex reassignment. The experiences of transsexuals show that some cultures still have work to do in becoming more accepting of diversity.) This is a task that we can begin to tackle now by fostering a climate of tolerance and acceptance towards those who are different from ourselves. We can also act to strengthen those institutions that prevent violence and protect human rights, for instance by building stable democratic traditions and constitutions and by expanding the rule of law to the international plane.

What about the hypothetical case in which someone intends to create, or turn themselves into, a being of so radically enhanced capacities that a single one or a small group of such individuals would be capable of taking over the planet? This is clearly not a situation that is likely to arise in the imminent future, but one can imagine that, perhaps in a few decades, the prospective creation of superintelligent machines could raise this kind of concern. The would-be creator of a new life form with such surpassing capabilities would have an obligation to ensure that the proposed being is free from psychopathic tendencies and, more generally, that it has humane inclinations. For example, a superintelligence should be built with a clear goal structure that has friendliness to humans as its top goal. Before running such a program, the builders of a superintelligence should be required to make a strong case that launching it would be safer than alternative courses of action.

References:

Yudkowsky, E. Creating Friendly AI: The Analysis and Design of Benevolent Goal Architectures. (2003, Version 1.0). http://www.singinst.org/CFAI/index.html