Algorithms, Data and Ethics
And you… you will always be remembered as The Architect that created this new world.”
“The architect?” I responded.
“The city of the future will not be designed by architects or urban planners.
We have already realized that no single human mind is capable of designing a new city. We have learned from our past mistakes.”
“I will be the trainer, if you will, the trainer of the algorithm that will automatically generate the most perfect civilization that has ever been envisioned.
We need to be objective, get rid of human biases, discrimination, and class division.
A human will not be able to do that but with machine learning and a set of guided algorithms we can rid ourselves of emotional constraints and design perfection”
Glòria Serra Coch. “The Noah’s Ark”. Assignment to design the city of the future. Theory of City Form (2018)
Urban planning, as a discipline has approached the design, planning and management of cities through different theoretical frameworks and tools. The fall of rational planning in the 1970s and the rise of discourses based on community engagement and acknowledgment of multiple subjectivities (Healey, 1992) brought as a consequence a distrust of the role of the expert and the “models” that he used to understand and shape the environment. As George Box stated 30 years ago, "All models are wrong, but some are useful. (Anderson, 2008)
Recently, however, we find ourselves being enthusiastically thrown against the miracles of data, which through mathematics and impossible to trace back algorithms, seem to allow us to take it one step further: we do not only completely rebuke the scientific method that rational planning relied on but we also do not even need to understand the processes at play to be able to make decisions (Anderson, 2008) . In this way, we can even overpass the subject (Amoore, 2011) and we can improve our efficiency by knowing for sure what people want without need to be losing time by asking them. “Petabytes allow us to say: "Correlation is enough." We can stop looking for models” (Anderson, 2008). Cities will not need to be designed by people, a self-feeding algorithm will decide what is the best for each area or group of people (Crichton, 2018). With systems like data derivatives, we can even annul the possibility of actual decision (Amoore, 2011).
However, are we, actually taking a step further or backwards? Are we, in fact, just substituting the causal relationships studied in the scientific method by pure probabilistic correlations that direct us in the “right” decision? Is this allowing us to impose an “expert” view of the matter that automatically erases any other types of associated knowledge? Aren’t we just replicating a top-down rational planning that merely substitutes the human “expert” by an AI and, in addition, allows us to eliminate accountability issues?