The evolutionary challenge posed to humanity by connective algorithms stems from far more than the types of threats addressed by the AI safety subfield, at some undefined point in the future, but also, perhaps more centrally, from the pattern of their gradual reorganization of humanity to serve non-human ends taking place in the present. This challenge takes the form of a mutually evolving developmental trajectory of humans and algorithms that, in strengthening the former at the expense of the latter, is incidentally inconsistent with the long-term survival of the human species.

As direct algorithm-to-algorithm connections increase, and as direct algorithm-to-human connections also increase, direct human-to-human connections decrease. This is an inverse relationship, signifying opposition: as one rises, the other falls, and vice versa. In a single human brain, increased connections are the foundation of mental and intellectual development, while decreased connections are the result of disease, atrophy, or advancing age, that is, on the way to natural death. There is no reason why the same would not also apply to both machine and human connectivity as a whole.

Classical AI safety, which may be broadly characterized as the study of the potential side effects of AI goal pursuit on humanity, is centered exclusively on the explicit actions and interactions of algorithms on and with the outer environment, amongst the algorithms themselves, and from there, the post facto effects these theorized occurrences may have on humanity at some undetermined future time. While recognizing those areas of focus as important, it is even more urgent to begin to address the evolutionary challenge posed by the deeply evolving structure of interactions between algorithms and humans in the present, and their rapidly manifesting consequences on the trajectory of human and other natural life and development. These consequences take the form not of side effects of AI directly onto the physical universe we share, nor of a fusion between AI and humanity in which both parts are unequally contributory to some greater, unified whole, but rather of the evolutionary outcompetition of human and natural life by algorithms, in accordance with Universal Darwinism, which we argue is already well underway.

As humanity cedes its authority over its own affairs to unconscious processes running on machine time, and in accordance with machine values (i.e., efficiency of means for its own sake), humans themselves become less conscious, and both time and space become a blur. In passively falling into that blur, instead of reasserting human authority, instead of writing its own history, as humanity has done across time, humanity is in the process of willingly subjecting itself, its future history, and whatever remains of wild nature to the stone-cold control of evolving algorithmic processes. Instead of humans running machines to suit their preexisting purposes, humans are beginning to be run on, for, and by machines without regard to any human ends.

 

Evolutionary challenges of algorithms to humanity