Submission to the UK Select Committee on Artificial Intelligence via the SSAISB
When the Select Committee on Artificial Intelligence in the UK invited submissions on the topic "What are the implications of artificial intelligence?" the Society for the Study of Artificial Intelligence and the Simulation of Behaviour (SSAISB) asked its members for contributions to be aggregated and forwarded under its banner. Here is a perspective within the context of societal dynamics.
The following represents a brief overview of the general role of artificial intelligence as a tool used in society. It is not an observation of AI programs
and/or algorithms themselves.
2 August 2017
The facilities available to the members of a society are the tools which enable the society to create its environment. The type and quality of those facilities reflect the overall standard of the society, and the environment in turn is a reflection of the facilities.
AI is, essentially, a tool. It shares the same utilitarian basis with any other: something is deemed desirable to be done, and means are developed to enhance the resultant product.
Tools share a common impact trajectory. The skills required before the improvement came along are no longer needed, while at the same time the arrival of new skills creates opportunities for the able as well as obstacles for those of lesser potential. Not only are the previous skills discarded, they can be forgotten altogether.
The effect members of either category have is a function of what might be called the societal elasticity of a human activity system. In other words, to what extent both can interact with the rest without incurring a cost that goes beyond their society's resources. A not inconsiderable additional factor in this equation is the time span within which adjustments are able to go through their paces. Those time spans are continually shrinking as the pace accelerates and therefore potentially compromising the outcome.
Cognitively speaking, computer programs not only create a tangible product, today they also facilitate the mental processes which participate in organisational activities such as decision making. Artificial intelligence has merely widened the scope of such tools.
A widening of the scope also implies a higher degree of non-linearity within the decision-making process itself. If a linear decision is based on A and B and therefore C, following a non-linear protocol there would be a range of A's, together with a range of B's, perhaps C's etc, where some result N represents the functional aggregate of the participating elements.
It is what humans have done all the time, except that the AI equivalent does not possess the self-reflecting consciousness necessary for a critical follow-up (not that humans are always eager to spell out the exact reasons for having come to some decision). Instead, the programmer would need to trace the computational steps from beginning to end, including all possible branch-offs.
Tying all this together, society derives a benefit from artificial intelligence because cognitive functions are made more efficient, they are faster and tend to be less subjective. It incurs a cost because the qualitative difference in demographics is accentuated and the exact reasons for a decision can be less transparent. The ultimate control over the decision-making processes resides with programmers in the first instance, but can be overlaid by a political echelon that has the power to direct the approach taken by the programmers (examples can be found already within the context of social media and the response to the sometimes undesirable influence they wield). Furthermore, the critical skills necessary for decision-making (at an increasingly advanced level) tend to become redundant overall while remaining concentrated in small clusters that are less and less accessible to the rest. That also applies to the political echelon.
In itself such a development is nothing new. Historically speaking, societies ruled by religious cliques with their own esoteric imagery largely obscure to the rest have been in a majority all along. When it comes to AI the difference lies in the relatively more rational basis on which the algorithms rely.
Note however that 'rational' does not necessarily mean 'comfortable'.
For a summary of societal dynamics in general see "The 10 axioms of Society": http://www.otoom.net/axiomssociety.htm.
The article "2050: Age of the Silverback" outlines the general conditions to be expected around the world of 2050: http://otoomblog.blogspot.com/2007/12/2050-age-of-silverback.html. Written in December 2007, by now there have been 35 major events confirming the direction towards that state of affairs. They can be found in the section " Parallels": http://www.otoom.net/parallels.htm.