In the future urban decision-making, should we listen to AIGC or people?

Source: Tencent Research Institute

Author: WeCityX

Content from: Tencent Research Institute & The Paper Research Institute & Long Ying Research Group, School of Architecture, Tsinghua University

Designing the cities of the future and enabling them to grow requires thinking about and understanding cities. We interviewed many of the leading thinkers and practitioners of the big questions about cities.

Technology needs to work for people. Scientific urban governance needs to be established on the basis of having enough data and a full understanding of the situation to better comply with the corresponding scientific and professional laws. It also needs to have forward-looking and leading thinking, and be able to Dial things into a healthy development track. The former point is what artificial intelligence machines are good at; but to achieve the latter point, it is necessary to rely on people to exert their own initiative in uncertain situations.

How to judge the current situation? Do you follow the direction indicated by the data, or do you rely on people to take the initiative to make decisions? Not only urban systems will encounter such propositions, people will face such important moments in their lives, and they may rely more on some kind of spontaneous instinct to make decisions. But the response of the urban system still depends on the existing mechanism, so it needs to be prepared before then. This may also be considered part of the emergency response.

Decision-making at critical moments

**Humans and machines have their own advantages, and in turn, they also have their own limitations. **In Chen Qiufan's view, the most fundamental problem of human decision-making is not that they cannot obtain comprehensive information, but that they are constrained by the limitations of their own consciousness. There may be many paths in front of us, but we can only see a very limited few - it may be the balance of interests, political conflicts, or historical burdens. It is very difficult to go beyond these limitations in consciousness.

**On the other hand, machine decision-making and human decision-making have completely different definitions in different scenarios. **Chen Qiufan gave an example, for example, under extreme circumstances, there are decision-making scenarios with high urgency, very serious potential consequences, and even large-scale wars. At this time, the proportion of people in it will be heavier.

Chen Qiufan said that the military departments of some countries have also had many discussions on such scenarios. For example, when drones perform military operations, they will use a lot of technical assistance, such as matching computer vision data, to determine whether it is an object that needs to be attacked. However, if the corresponding technology is used in weapons of mass destruction, the potential risk of accidental injury will become very large. It is equivalent to handing over the lives of many people to the hands of machines. Therefore, there is a principle of "keep human in the loop", that is, when deciding whether to execute the attack command in the final stage, someone needs to make a decision.

In contrast, in less urgent and important daily life scenes, people need to retreat relatively, and machines have to play a greater role. Because machines are better at reading data and recognizing patterns. At this time, the role of people is equivalent to gatekeeping procedural justice.

However, Chen Qiufan also realized that in the daily state, there are some things worth worrying about in the decision-making process of cooperation between machines and humans. Because the system is not static, but in the process of evolution and feedback. In other words, people will also be trained and changed. Chen Qiufan pointed out that in such a system, if enough tasks are performed, people will tend to become more like machines—whether they are domesticated by machines or accustomed to handing over more decisions to machines. This is another feedback loop.

Perhaps, people cannot always make the right decisions, but they always need to have the awareness of being responsible for themselves. Chen Qiufan said that in this sense, we have returned to a very simple and humanitarian concept. It's a very big, but also very small realization that each of us is responsible.

Human decision-making can tolerate chance

According to Raz, deputy editor-in-chief of Science Fiction World Magazine, most of the current urban life is spent listening to the sound of machines. For example, in the current travel, most people follow the machine navigation completely, and such operation usually does not cause any problems. But he also pointed out that the most sensitive thing in the government is that it involves large-scale public services and the specific safety of everyone, and it needs to be gradually improved and connected. **

In Raz's view, artificial intelligence must be used in certain conditions. ** Once out of this type of scene, it cannot be used. People make programs that deal with different scenarios. A large number of problems arise in the link of docking and association. In other words, when compatibility is required, the strengths of artificial intelligence are not easy to play. For example, Raz said that in terms of traffic, an artificial intelligence dedicated to traffic light control and another artificial intelligence dedicated to lane GPS positioning will have weak connections and crossovers. This requires coordination between departments and groups.

In addition to traffic, there are many more complex scenes in the city, and the corresponding problems are more difficult. In other words, for cities, it is necessary to allow point-shaped artificial intelligence to better interact and connect them to each other, so as to better reflect the situation in a timely manner and support more scientific and complete decision-making.

But machines support decision-making, and cannot and cannot exhaust all aspects of the city. Raz also said that if it is a machine decision, it will be like a kind of DNA replication that is completely infallible. For organisms, in the process of DNA replication, genetic mutations are required to achieve the progress of species. There must be something wrong with this. Perhaps, from the point of view of individual genes, this mutation is meaningless, but it is potentially beneficial to the overall evolution. Only then does this system have endless possibilities.

City life is similar. Raz said that human beings have a side that pursues curiosity and will do things that are just for entertainment and fun. These things seem meaningless, but they are very valuable to the overall evolution.

These complex parts cannot be realized by machines. In the words of Wu Tinghai, a professor at the School of Architecture of Tsinghua University, the complexity created by technology is still simple complexity, and it is difficult to break through the real complexity. **So machines can assist humans, but cannot dominate humans yet. **Wu Tinghai is cautiously optimistic, and believes that overall, the changes brought about by technology are currently predictable. In his view, new technologies bring about new ideological changes, which is also a manifestation of innovation. On the one hand, "changes lead to success, and changes lead to long-term", change is the normal state of development. On the other hand, "keeping the center" is also a kind of high rationality, "center" represents people, and technology is still serving people.

Back to the value ethics of technology

Zhang Yuxing, a researcher at the School of Architecture and Urban Planning of Shenzhen University, also saw this aspect. He pointed out that we should think about how to defuse the possibility of technology being used as a control tool. This is because some existing technologies may actually reinforce inequity. At the same time, those with Internet tools have greater control. This is also the concern of many Western philosophers of science and technology. Technological systems bring about equality of human rights, but the problem of inequity between those in control of the system and others remains. This should become a principled control issue for ICT reverse consideration. On the other hand, we should think positively about how to make people's control over resources and energy develop on a smaller and deeper scale to alleviate inequality.

Zhou Rong, an associate professor at the School of Architecture of Tsinghua University, gave us more warnings. At the macro level, he pointed out that technology cannot be denied, but technology must be domesticated by humanity before it can be better used. At present, ubiquitous anxiety, introversion, oppression, ignorance of meaning, and lack of sense of value are all symptoms of the Internet age, reflecting the structural problems of the current civilization in the process of transformation. In fact, technological progress may bring about another civilization. The construction of a new civilization cannot be dominated by technology, and someone must be responsible for it. This needs to go back to the level of value ethics to construct.

Zhou Rong said that technology is the material for organizing civilization, and it will give birth to a new civilization at the moment. The existing ideological propositions all emerged from the old civilization. Therefore, it may be more important to maintain critical thinking.

WeCityX team summary:

**The rapid development of artificial intelligence and the replacement of human beings in various fields are often worrying and even frightening. How to make full use of machine intelligence, and give full play to the value of people, and cooperate well in various scenarios? **

**Hybrid intelligence, or human-computer hybrid decision-making, may be a more realistic approach at present. Its two manifestations are "hybrid augmented intelligence based on cognitive computing" and "hybrid augmented intelligence with human in the loop". The former refers to the introduction of various biologically inspired computational intelligence technologies into the intelligent system, such as fuzzy systems, neural network systems, and evolutionary genetic algorithms. The latter directly introduces the role of people into the intelligent system, forming a human-in-the-loop or even a human-centered hybrid intelligence. The former is good at making quick decisions at critical moments; while the latter may be the most important feature of the new form of human civilization. **