Paper short abstract:

If a driver swerves to miss a pedestrian, who is responsible if the computer control of the cornering radius does not allow the vehicle to use the utmost potential of the tyre traction? What legislation governs the control system? Can grace, anthropology and engineering mesh to manage these issues?

Paper long abstract:

James Clerk Maxwell died in 1879. In his 48 years he uncovered the phenomenon of electromagnetism and set in place the mathematics of closed loop feedback control systems. Ninety years later, engineering students fed punch-cards coded with “if statements and conditional do loops” into enormous computing machines out of which came lists with data segregated into categories. That was not machine learning. An intelligent machine, senses data and segregates it in a manner not predetermined by the programmer, but correlated within the machine itself with the most significant variables being sensed. Thus a vehicle cornering radius may come to be adjusted not according to a fixed algorithm but by an algorithm developed within the vehicle’s own computer. Such implications find resonance in the work of Dame Mary Douglas on risk, danger, blame and culture, Edwards Deming’s concept of profound knowledge and Daniel Kahneman’s revelation of human overeagerness to simplify without understanding. Regulations surrounding machine learning must heed erstwhile Chief Justice of England Lord Hewart's warning of “The New Despotism” in which the executive holds the legislature in its thrall, arranges the passing of the Acts it administers thereby shielding itself from the judiciary. Technical complexity makes this all too easy and the legislature all too gullible as guidelines and standards morph into regulations that become treated as immutable maxims. This paper explores Hewart’s despotism as a core anthropological concept and the form of grace required to hold on track Edmund Leach’s and Anthony Giddens' runaway worlds.