Artificial intelligence is one of the great promises of today’s technology. It is associated with the megatrend of digitalisation. At VTT, we set out to examine the relationship between this promise and urbanisation, another key megatrend shaping our future. We focused on where the two megatrends – digitalisation and urbanisation – converge, i.e., the smart city.
On 17 November 2017 we held a My Smart City workshop, where we asked what kind of opportunities do artificial intelligence and augmented reality offer for developing cities. What kind of data do cities have and how is it being used? What kind of future visions are there, and how could artificial intelligence and augmented reality contribute to joint development of cities? At the end of the workshop, we addressed ethical and psychological themes surrounding the subject.
In this blog, I will focus on artificial intelligence (AI), as it seems to provoke more contradictions and, at the same time, to provide new opportunities in the development of cities. Augmented reality (AR) is already a more widely approved phenomenon. AR can be used for visualising future solutions in urban environments for city dwellers and other decision-makers. This is a fantastic and useful thing, and, at the same time, it is great fun and fascinating to step into the future by wearing virtual glasses! However, in this context the opportunities offered by AI fascinate us even more, as they give people something to object to as well.
Having AI as a mayor?
The planning of city operations and environments is filled with decision-making that shapes the future. In a good decision-making process, information is first collected and analysed, and then a decision is made. Could AI handle this process, and perhaps even be better in it than, for example, a politician? AI can collect data automatically, and it understands the present situation on the basis of a large amount of source material. After this it can predict the future, and, on top of that, gradually learn more in the process. Someone has even suggested that AI could serve as a mayor for cities! It has also been said that AI is a way of keeping people far away from decision-making for as long as possible, and this way better decisions can be made without all kinds of appealing parties slowing down or disturbing the decision-making process.
But how could we trust a decision-maker that cannot distinguish a chihuahua from a blueberry muffin? Artificial intelligence cannot necessarily make this distinction; and even if it learned to do so little by little, in practice AI can only compare data with previous images, after being told what they represent. In other words, after seeing enough muffins and chihuahuas, it can begin to distinguish between them with reasonable certainty. AI therefore learns linearly. In that sense, it also sees the future as a continuum of the present. How could it handle societal transformations in which the direction of development changes?
AI always traces its origins in a person, a person who has his or her own values, culture, interests and understanding. Because AI is created by people, it is culturally embedded. Furthermore, the material collected by the machine, the analysis and decision-making reflect certain values and beliefs, in the same way as they would reflect human ones. If we want AI to analyse for us how effectively we could create a residential area with 25,000 inhabitants and 10,000 jobs, we teach it to make such analyses. Instead, if we wanted to have information on what would make 25,000 people satisfied with their residential area, we would need different starting points for the analysis.
In my personal opinion, the great potential lies in the discussing AI. We could, for example, install discussing benches or swings in parks or playgrounds, allowing us to collect information for decision-makers based on discussions between AI and citizens, while simultaneously creating an amusing discussion experience for citizens. At least my 8-year-old son likes to talk to and amuse himself by having conversations with the mobile phone AI. Someone in the park who sits down on a bench to rest for a while could either talk with the bench, if he or she so wishes, or tell it that I’m sorry, but today I don’t feel like talking.
In my opinion, another promising observation of the opportunities offered by AI is a joint intelligence, where an entity consisting of a human being and AI is capable of making better decisions than earlier. One example of this could be a process, where AI is used for collecting data and making a rough analysis of it. The results of this rough analysis would then be given to people to process, allowing their experience and everyday understanding to affect the analysis. Basically, we are already implementing such a process when using, say, a navigator. If we have earlier experience-based information on the route being selected, we may even select a route opposed to the one suggested by the navigator. In other words, we do not allow the navigator AI to guide us when we do not need any guidance.
However, large masses of data are often impossible to process logically by human brain; here, the AI can be of assistance. As suggested by AI researcher Timo Honkela, we can also believe that we could enhance democracy in decision-making by using AI; the use of AI would allow thousands or even a higher number of people to participate in the decision-making process for real. In other words, AI could really be an integral element of a smart city, a supportive tool for decision-making. I still keep on hoping that this kind of development would become stronger, but in such a way that we, the people, stay in control of how things are managed.
Research Team Leader, VTT
This blog is the first in our series of blogs focusing on artificial intelligence and augmented reality in the operating environment of smart cities. The next articles in the series will be published under VTT blogs during the early months of 2018. With this series of blogs, we wish to enliven the debate about artificial intelligence and augmented reality.