Our central claim is that, despite the conclusion of the previous section, there is no reason to view the possibility of a robot takover as, in any way, inevitable. The writers cited tend to conflate the issue of inevitability with that of possibility, but this seems highly mistaken. We will not here consider the level of technical probability involved, nor the time scales used in the predictions of Moravec, de Garis and Warwick. We will pass over these contentious issues with the simple remark that predicting the future of technological developments has always been extremely difficult.
Our main counter-argument to the various predictions of a robot takeover is that they seem to ignore the realities of the application of power in human societies. Taking Warwick first, and his account is in many ways the fullest and most coherent, he repeatedly asserts that the possession of intelligence is sufficient to gain power. This is clearly false. Even if we allow Warwick to supplement pure intelligence with real-world knowledge, which is, as Perri 6 points out, extremely expensive to acquire, there is still no obvious correlation with power. Human societies have not been, and are not, ruled by the most intelligent or the most knowledgeable. Other factors are at least as (and probably more) important in gaining power in human societies.
It is highly simplistic to make political power, or the potential for power, equivalent to intelligence or knowledge or other measures of cognitive capacity. While this may be understandable hubris in academics and intellectuals, it is not obviously supported by any analysis of political history. The processes by which power moves between groups and individuals are extremely complex and beyond the scope of this paper. However, there is no support for the simplistic assumption that mere intelligence in robots will correlate with power in human society (note3).
Returning to the claims made by Perri 6 we can agree with him that the acquistion of power requires far more of machines than mere intelligence. The most important omission by almost all writers is that of motivation. It is unreflectively assumed that machines, should they acquire the ability to dominate humanity, would inevitably be motivated to do so. This may well be nothing more a piece of subconcious projection of human motives. If there is any discussion of the motives for a robot takeover it is subsumed under the claim that the machines would know best (in Warwick's prediction) or under the motives of pro-robot humans in de Garis' prediction. It seems far more likely that the pursuit of power over humans would require a great deal of effort by the robots, particularly, as observed by Perri 6 in the acquistion of real-world knowledge relevant to the task. We are entitled to wonder why the robots would engage in such an effort.
What is important to stress here is that there is no inevitability about any future machine being motivated to acquire domination. We are hardly in a position at the present level of technological development to initiate discussion on the goals of any intelligent (or even sub-intelligent) artifacts. However, we assert strongly that there is no reason at all to suppose that the motivation to dominate humans would have be programmed into any conceivable future robots or that they would automatically acquire such motivation. It is, perhaps, conceivable that such motivation could be deliberately programmed into robots by malevolent humans. However the problem of human malevolence seems both familiar and highly distinct from the problem of a robot takeover.
In fairness to Warwick it must be observed that he stresses the fact that 'intelligence' is not easily defined and that direct comparisons of human and machine intelligence are very difficult. Nonetheless his conviction that machines will come to exert tyrannical domination over humans stems primarily from his claim that the machines will soon possess vastly superior intelligence. No factor other than intelligence is cited in support of the tendency towards tyrannical domination.