Post-algorithmic smart cities

Mona Roman, director of innovation at Metropolia University of Applied Sciences, Helsinki, wrote a review about our book Smart Cities in the Post-Algorithmic Era: Integrating Technologies, Platforms and Governance published by Edward Elgar. On this occasion, this note refers to the title of the book.

Smart Cities in the Post-Algorithmic Era is a collection of papers that discuss technologies, platforms, and governance for making cities smarter, more effective, efficient and innovative. The main argument is that algorithmic logic is not sufficient to produce truly smart cities. Public authorities and organisations in cities should develop solutions that go beyond algorithmic systems. Moreover, in developing innovations for cities, non-algorithmic systems are more important rather than algorithmic ones.

An algorithm is a process for solving a problem in a step-by-step procedure that concludes with an outcome. If a procedure is not conclusive, it is not algorithmic. Non-algorithmic is a process not involving an algorithm or not capable of being expressed using an algorithm.  In cities, many processes are algorithmic. On the contrary, those that do not rely on well-defined routines are non-algorithmic. All innovations are non-algorithmic for evident reasons. They are produced out of the box, out of standard procedures. Also, there are many computer programs and applications that are non-algorithmic. While every application includes a list of instructions, this does not make it algorithmic. Non-algorithmic applications allow users to make decisions and determine the outcome. The MS office is a good example; while the PPP enables the making of presentations, the user gives the instructions for the design of a presentation, not the program. A PowerPoint presentation is the outcome of human intelligence supported by a computer program.

Before 2010 most smart city applications were non-algorithmic, guided by web-based instructions given by the user. Even more, in solutions based on crowdsourcing platforms, the outcome depends on choices made by a hundred users. With the excessive use of sensors and actuators in transport and utility networks the landscape changed, and smart city solutions started being guided by algorithms. Moreover, in applications based on AI, the user could define some goal and the application would make all calculations for the best prediction to the given goal. The recent successes of artificial intelligence created a vision of smart cities heavily relying on AI and choices made by machines.

However, a fully algorithmic city is not a good urban system. Cities are places of creativity and innovation. To sustain these functions, smart cities should remain non-algorithmic, enabling innovations. In an algorithmic context, innovation is zeroed. New solutions for growth, sustainability and safety cannot be produced by machines alone. This is the central argument of Smart Cities in the Post-Algorithmic Era. Smart cities emerge from collaboration technologies, digital platforms, Internet of Things, social media, data science, and AI. The algorithmic logic, under which these technologies operate, can be much more effective if combined with other sources of intelligence available in cities, such as human intelligence, creativity and innovation, collective and collaborative intelligence. The concept of ‘post-algorithmic cities’ refers to this combination of machine, human, and collective intelligence in smart cities.

The algorithmic logic is dominant in many applications for smart cities that address traffic congestion, find parking spaces, use car-sharing, detect and quickly respond to accidents, optimise the allocation of resources, promote buildings and utility efficiency, energy-saving and the use of green energy, forecast environmental conditions. Currently, the most advanced algorithmic logic can be found in deep learning solutions that allow machines to learn from data. It is a fascinating idea, a machine that can learn by supervised learning methods and reinforced learning algorithms. This is a different approach from traditional AI, which focused on the processing of symbols according to rules, reasoning, and transformation of symbols through programming. This type of AI, based on learning and training from data, has proved very effective in problems related to computer vision, speech recognition, natural language understanding, robotics, medical image analysis, and self-driving vehicles, which feed various smart city solutions.

But, when we move to more complex urban problems, for instance, challenges of growth and sustainability, which are distributed and ill-defined problems, with models changing quickly over cycles and solutions depending on institutions that are not transferable from one place to another, both traditional and current AI are less effective. In problems heavily depending on the context, such as those of growth, government, and poverty, machine intelligence is not enough and should be combined with other sources of intelligence – human and organisational intelligence – to produce effective solutions. Thus, smart city solutions should stand on a wider base of intelligence, combining machine intelligence with human intelligence provided by human agents and collective intelligence available in human communities and institutions. Bringing those sources of intelligence together and finding solutions to how to do so is a big challenge of our era. A hint on the direction of research toward this goal is the subtitle of the book “integrating technologies, platforms, and governance”.

Smart Cities in the Post-Algorithmic Era: Integrating Technologies, Platforms and Governance, edited by Nicos Komninos and Christina Kakderi (2019), Edward Elgar, Cheltenham, ISBN 978 1 78990 704 9

A review by Mona Roman: Smart Cities in the Post-Algorithmic Era – Book review