Cloud Infrastructure for Health in Low and Middle-Income Countries: Becoming Increasingly 'Autonomous'?
Way back in the 1970s, Langdon Winner coined the term ‘autonomous technology’ to theorize the relation between technological complexity and control (Winner 1978). As technologies become increasingly complex, and as the locus of control shifts to non-local locations, technology becomes increasingly ‘autonomous’, taking on a life of its own and thus becoming more difficult to control. As an example, the blinds of the windows in the modern office at the University of Oslo in Norway are controlled by automatic sensors. Depending on the intensity of the sun (or its absence), the movement of the blinds is auto- controlled. From the perspective of an occupant of the office, the technology is becoming increasingly autonomous giving him or her limited control, as compared to earlier when the window could be just manually opened or closed, or the curtains pulled out or in. Now the movement of the blinds is under increasing control of non-local forces, with decreased control of the local actors. The theme of ‘autonomous technology’ has also been picked up by science fiction novels and characters. Frankenstein was perhaps the first of these, representing a creature built by technology turning against its very creator. Now movies and television shows are full of run-away artificial intelligence systems where the robots become autonomous and seek to control and destroy the masters who created them. The Matrix trilogy and I Robot are perhaps the most successful cinematic examples of this trend in science fiction.
In a much more mundane manner, we see problems of autonomous technology in daily life. An interesting question this raises in the context of HIS in LMICs is whether the cloud and big data have tendencies to take on similar autonomous characteristics, viewed from the perspective of the user or the state in the LMIC, and if so, what are its implications? The technologies of the cloud, big data, and associated analytics are in themselves complex, when combined more so, and are normally under control of multiple non-local agents.
In the following section, we seek to understand the alternative models being used to operationalize these technologies in LMICs.