全球汇 | 伊丽莎白·席尔瓦访谈：复杂性理论、“软”人工智能与当代中国城市发展
采访：刘伦，剑桥大学土地经济系博士研究生（L）
L: We see that you are also among the earliest urban researchers to write about complexity and planning, manifested by your book ‘A Planner’s Encounter with Complexity’. Could you briefly introduce the major difference between understanding planning from a complexity perspective and from traditional perspectives?
我们了解到，您是最早就规划与复杂性发表论述的城市研究者之一，以您2010年出版的著作《规划师与复杂性的相遇》为代表。能否请您简要介绍，复杂性视角与传统的理解规划的视角有何差别？
S: Certainty was never a fact of life and shouldn't be a presumption for planning, because planning is based on scenarios of the future, and the future is not made of certainties, but discoveries and changes. The planning system could not produce what it used to promise. Around 20-25 years ago, most of the planning systems around the world were still based on certainties. The decision makers wanted to know exactly what was the result of a planning exercise and would feel safe if a rational road map was presented in order to implement it, just then to get very disappointed. But theories and practice needed to evolve, we didn’t know very well how to make things different. Most of us knew that planning wasn't working when we gave 100% certainties - I realized that it was actually possible to move from certainties to probabilities, and when you do so, you enter the world of complexity. Because, if the world is not one plus one equal to two, then you need to understand how complex relations happen in the physical and social world, and then adjust. You need to understand causalities and impacts，evolution and feedback loops, and that it takes different lengths of time for things to happen. All of them above are complexity. We raised this issue in the AESOP conference in France. We then linked the experts in different areas in order to produce that book.
确定性从来不是真实的生活，也不应被作为规划的前提，因为规划是基于未来的场景，而未来不是必然的，它是基于不断的发现和变化。我们的规划体系无法再实现那些它过去基于确定性的承诺。大约在20~25年前，全世界大多数规划体系都是基于确定性的。决策者们想要确切的知道一项规划措施的结果到底是什么，比如希望看到一张即将一步一步实现的路网规划图，但他们最终往往会失望。但是相关理论和实践当时还没有获得充分的发展，所以我们不太清楚如何去改变。但大多数人都知道，规划不会给出百分之百确定的答案。于是我意识到，或许的确可以从确定性转入可能性，而这样就会进入复杂性。如果这个世界不是简单地一加一等于二，你就必须理解复杂关系如何在物质世界和社会世界中发生，并作出相应的调整。你要理解因果性和影响、演化进程和反馈回路、以及不同的改变需要不同的时间，这些都属于复杂性。在法国召开的欧洲规划院校组织（AESOP）会议上，我们首次提出了这个问题，会后，参会不同领域专家合作出版了这本书。
L: It seems that the complexity approach is more of a research paradigm instead of a practical paradigm, are there any example in the planning practice in the UK or continental Europe that reflects the idea of complexity?
从复杂性出发的方法似乎更多的是一种研究范式，而非实践范式，那么在英国或者欧洲大陆有没有反映了复杂性理念的规划实践案例呢？
S: I think since the sub-prime crisis, politicians around the world have realized that the world is not certain, and these shocks can be very real. Yet although decision makers understand that uncertainty is important, it’s still difficult to implement it in planning agendas at the policy level. The politicians in their 60s were raised as leaders in a rational world, so it’s very difficult for them to understand the mechanics of complexity theory. Thus they will be very uneased and suspicious about uncertainty, even if they know that we live in an uncertain world. Sometimes it is like they have two discourses: they acknowledge the importance of uncertainty, but then they ask for things that unless you have a rational model you cannot produce.
我感到自从次贷危机以来，全世界的政治家都认识到这个世界并非一成不变，而这些打击真的可能在现实中发生。决策者们知道不确定性是很重要的，但很难在政策层面上把它应用到规划中。这些六十几岁的政治家们是在一个理性世界中成长起来的领导者，因此他们很难理解复杂理论的机制。即便他们知道这个世界是不确定的，他们依然会对不确定性感到疑惑不安。有时他们会自相矛盾，他们一方面承认不确定性的重要性，而同时又要求得到只能由完全理性模型给出的确定性的答案。
If the last 20 years of researchers didn’t have done a lot of work, we couldn't understand complexity theory. Most researchers, don't see the problems of the past as mistakes from past models of the 1950s, but just saw an opportunity to change, build, to expand, to enhance, or to create new innovations. We academics don’t think that something that does not work is bad, we implemented because we assumed it was good and as a good researcher we say "didn't work well, what can we do in order to improve", but in the real world, failures are things that don’t work and end up producing very bad consequences. I have been arguing that the best way to implement complexity theory in the real word is by using 'adaptive planning policy and practice'.
如果过去20年的研究者们没有坚持下来做大量的工作，就不会有今天的这些对复杂性的研究成果。大部分研究者不认为1950年代被大量批评的模型是错误，他们从中看到了新的机遇，去建造、扩展、提升、创造的机遇。在学术界，不成功的东西不一定是糟糕的，一个好的学者总是在考虑如何改进。但在现实世界中，如果某些事物没有成功，而且还产生了糟糕的后果，它们就是失败的。也因此，我一直认为，在现实世界中实现复杂理论思想的最好方法是借助适应性规划。
L: Actually, ‘interdisciplinary’ has been one of the major topics in urban planning and research for quite some time. How do you think about the progresses made by the academia in this area by far? What do you think are the major gaps and major challenges?
事实上，在相当长的一段时间中，“跨学科”都是城市规划和研究中的主要问题之一。您如何评价学术界迄今为止在这个领域中取得的进展？您认为其中最大的问题和挑战是什么？
S: That is a very good question, which I will answer very carefully depending on the context. If someone asks me that question in the US, my answer will be different from that in Europe. Europe in the last 20 or 30 years emphasized a lot on qualitative research, not quantitative. So, Europe evolved a lot in participative planning, in questionnaire design, etc., driven by the great work by people like Pasty Healey. But unfortunately we need to move the quantitative side too, simultaneously. US has a lot of quantitative, but they should have more qualitative. Most of the planning students in spatial planning in US know about GIS, while in Europe, no. So you can have in Europe students in planning with no spatial awareness. Europe moved forward in qualitative studies, and that was good, but now ends up forgetting the quantitative analysis in planning. In UK and continental Europe, the amount of money to do quantitative research in planning once reduced. As a result, the courses also ended up reduced. So that’s why we have very strong groups but not very many. The US did it a little bit the other way around: moved very fast in quantitative, but some universities either have no qualitative research or waste a lot of money on mostly quantitative research. I think in China the reality is a bit different. When I speak with people in China, they are very aware of the importance to link qualitative research with quantitative. The hope is that now they have a good balance between qualitative and quantitative, and don’t repeat the mistake of just funding one part without funding the other.
这个问题非常好，我必须非常谨慎地根据具体情境作答。如果有人在美国问我这个问题，我的回答恐怕就会和在欧洲不同。在过去的二三十年中，欧洲非常强调定性而非定量研究。以帕齐·希利（Pasty Healey）为代表的学者使欧洲在参与式规划、民意调查等方面取得了很大进展。但遗憾的是，我们的定量研究亟待推进。美国有大量的定量研究，但定性研究需要补充。大多数美国的规划学生都会用GIS，但是在欧洲很多规划学生不会，有些学生可能毫无空间意识。欧洲在定性研究中大踏步前进，这很好，但却忽略了规划中的定量分析。在英国和欧洲大陆国家，定量规划研究的经费一度被削减，相关的课程也随之减少。这就是为什么欧洲有很棒的偏定量的规划研究团队，但是总体数量不多。美国则刚好相反，在定量领域发展迅速，但一些大学完全没有定性研究，或者将过多的经费浪费在了定量研究上。而我认为中国的情况稍有不同，我在和中国研究人员的交谈中发现，他们非常了解联系定性与定量研究的重要性。希望他们能够平衡好定性和定量研究的关系，不要再顾此失彼。
I think this differencecan be explained with an historical view. Except during the first decades of the XX century and the post second world war, Continental Europe was always influenced by the French school and the Dutch school of planning,which are very design-oriented.Whilethe Anglo-Saxonic world is more quantitative-oriented, I think now we have more integration again. With our technology evolved, in the future we will have hybrid approaches to planning where you have design and modeling happening at the same time. Most schools of architecture and landscape planning in Continental Europe already link the design-based to modeling approaches, like in Vienna University and in Cologne University.
我觉得这种区别可以从历史的视角来解释。除了20世纪初和二战后的一段时期，欧洲大陆一直以来都受到法国规划学派和荷兰规划学派的影响，这两个学派很大程度上是以设计为导向的。而英美国家则更具定量导向。我认为如今这两种类型具有更多融合的趋势，且随着科技的发展，在未来我们会见到更多设计和建模的“合体”。如今欧洲大陆大多数的建筑和景观规划学校已经在将设计导向与定量建模相结合，比如维也纳大学和科隆大学。
And compare it with the US: the Harvard design school is very design-based, but MIT is very modeling and quantitative, so it’s very difficult to say. My experience says that in most countries that are growing very fast,such asthose in Asia, South America and Africa,people first implement the traditional Anglo-Saxonic tradition of master planning, the rational planning. As they become more relaxed from the pressure of development, they take more incremental and comprehensive approaches and putmore emphasis on design. But a lot of the design-based now is also computational, so you don’t see the separation between quantitative/qualitative that you would see 20 years ago.
美国内部也存在这样的差异，哈佛的设计学院是非常强调设计导向的，而MIT则很强调建模和定量分析，所以也不能一概而论。以我的经验来看，在快速发展的国家，如很多亚洲、南美洲和非洲国家，人们会先采用传统的“英美传统”的总体规划、理性规划。随着发展的压力减小，人们会逐渐采用渐进式的、综合的规划方法，对设计的重视程度也越来越高。但现在许多设计主导的规划方式也大量依赖计算机，所以这种定性定量的分野不会像20年前那么明显。
L: As early as ten years ago, you were already writing about artificial intelligence and planning. Now as there is eye-catching progress of machine learning and other AI technologies, especially the recent victory of AlphaGo over human champion, the potential of AI in urban planning is becoming a hot topic. What do you think about this? Do you think it is merely a ‘fashion’ trend for the field of urban planning/research, or it is going to make substantial changes to how we plan and study the city?
早在10年前，你就写过关于人工智能和规划的文章。如今机器学习和其他人工智能技术都取得了夺人眼目的进展，尤其是最近AlphaGo打败了人类冠军棋手，因而人工智能在城市规划中的可能性也正成为一个热门话题。您如何评价这些现象？在您看来，人工智能在城市规划和研究领域是一种“时髦”潮流，还是会令我们规划和研究城市的方式产生实质性的改变？
S: I think the words, as any other words, will come and go, but the impact will stay. As to artificial intelligence, we need to be very careful how we call it. In planning, I think it is better to call it ‘soft artificial intelligence’, ‘soft AI’, because in our systems there is a level of calibration, so it is very difficult to say that you just have a machine thinking like a human being. We always require some level of human intervention, and this happens at all levels, i.e. stakeholders are human beings, the planning process will always be enhanced if there is human intervention (we can't go completely, 100% artificial). But nowadays, most of the things you do already has a little bit of AI. For instance, in global cities, traffic lights have two softwares one of them prepared for unusual circumstances that activates an AI. When there is an incident, the traffic lights will change locally according to the soft AI algorithm, that is machine learning. Though most of us are not aware, artificial intelligence is already existent on a regular basis.
我认为这些说法，和许多类似的说法一样，都只是一时的，但这些事件的影响是长远的。至于人工智能，我们必须对它的命名很谨慎。在规划中，我觉得叫它“软”人工智能会更加合适，因为我们的系统中有相当多的校准工作，所以很难说它是个能像人类一样思考的机器，我们永远都需要一定程度的人工干预，这种干预发生在各个层面上，包括利益相关者、规划过程等。但如今许多事都已经有一点人工智能了，例如全球性城市的交通信号灯有两套软件，其中一套会在特殊情况下启动人工智能，如果某地发生了事故，这一地区的交通信号灯就会通过机器学习，或者说“软”人工智能算法来做出相应调整。人工智能已经来到我们的生活中了，只是大多数人都还没意识到。
But AI is not as new as that. Crowd sourcing software uses cameras that detect movement of people, and what Alan Turing started to develop in terms of identifying patterns is another good example. It highlights patterns of clustering so that stampedes and other dangers can be avoided. Most mega-events already have an artificial intelligence software like crowd sourcing, which utilizes image recognition to save lives. So there are things happening, most of which include some sort of soft artificial intelligence, but I think 'soft-AI' human intervention is and will always be important.
但人工智能并不是全新的事物。人群监控软件使用相机来测查人的移动，并从中识别模式，它使用的算法可以一直追溯到阿兰•图灵（Alan Turing）。它会将人群聚集的模式高亮显示，这样踩踏等事故就能得到避免。大多数超大型活动都已经用上了像人群监测软件这样的人工智能，利用图像识别来保障生命安全。这方面已经有很多进展，其中大部分都是“软”人工智能，它们正在也将继续发挥重要作用。
L: What are the biggest challenges of incorporating AI into planning?
要将人工智能融入规划，最大的挑战是什么？
S: Getting too excited over artificial intelligenceis a risk and a mistake. We need to keep focus in what is the goal of planning and why we use computers (to create better societies and help human beings). Someone invented these nice, popular words, but they need to be very careful because while in other subjects such as physics, in material science and computer science, it is possible to go 100% AI. We use computers to better plan, but it will always be very difficult to separate the human factor from this process. In the traffic lights example, when an incident happens, you can calibrate the traffic lights. There is a very high level of automation in cases like this, where you don’t need human intervention on a regular basis. But the notion that we can just relax and that artificial intelligence will plan our cities and manage our lives is not going to happen,even without the calibration. Reality changes so fast that it will be extremely risky to leave it only to artificial intelligence.
对人工智能热情过度是有风险的，也是错误的。我们应始终明确规划的目标和我们采用计算机技术的原因（改善社会和人类生活）。有人发明了这些令人动心的流行词汇，但要非常谨慎，因为我们在做的事情不同于材料科学、计算机科学等，这些学科有可能在一些情况下百分之百借助人工智能。我们的确可以利用电脑来更好地进行规划，但很难将人的因素彻底剥离出来。在交通信号灯的例子中，信号灯在事故发生时进行自动校正，这种情况下的自动化程度可以很高，不需要时常进行人工干预。但是这个案例在一定程度上属于特例，我们不可能彻底放手，让人工智能来规划我们的城市，管理我们的生活。现实变化如此之快，完全将它交给人工智能掌控是极其有风险的。
L: What do you think will be hot topics in interdisciplinary planning/research in five to ten years’ time?
您认为在未来的5~10年中，跨学科规划和研究会有哪些热门问题？
S: I think in the next five years it is going to be full integration. Not only integration methodologies between sectors, but in terms of computer models. We now have people developing different computer models, but we still don’t have a platform that mixes computer models together. In the future, it’s going to be very easy to integrate modeling languages and modeling platforms so as to couple multiple models. Moreover, it is very easy now to build new models in softwares like Netlogo, which is going to be even easier in the next 5 years. Planning schools will also start to have courses on these subjects, because at this moment, most of planning schools only teaches GIS, and not complexity or adaptive planning. That is a problem that I think will be solved in the next 5 years.
我认为未来五年中，研究的“整合”将是一个重点，不仅是之前提到的那种各部门规划之间的整合，而且是对分析模型的整合。现在很多研究人员在开发不同的计算模型，但我们仍然无法在一个平台上将它们都整合起来。未来这样的做法有望变得非常轻松，整合不同的模型语言和建模平台，从而耦合多个模型。另外，如今在Netlogo等软件中建模已经比较简单了，在未来五年会变得更加便捷。而且规划院校也会开设更多这些方面的课程——当前大多数规划院校还只教授GIS，不教授复杂性和适应性规划相关的内容。我认为这些问题会在未来五年中得到解决。
In the next 20 years, I think the problem will be inanother area, whichis the impact of new technologies on our lifestyle and the way we understand urban issues. What if we create an energy supply that is so cheap that distance costs become no longer a problem? There will be more cars, and cities don’t need to be compact - they can sprawl. Then even if you want to do mass transportation, more people will prefer private transportation. It is very good for the city to change the type of energy that we have in cars from polluting to non-polluting. But, how will the city management cope with that if the new energy is so inexpensive that the cost function that keeps cities compact ends up disappearing?
另外，未来20年中我们面临的主要课题之一将是，科技发展对城市生活方式的改变和我们对各种城市问题态度的改变。比如，如果我们发明一种廉价的能源供应方式，让汽车行驶的成本、能耗和污染不再成为问题，那么可能公共交通很难再具有吸引力，而且我们是否还有必要坚持高密度的城市发展模式？城市管理者要如何应对这些变化？
L: In the end, as a scholar with rich research and practical experience in both Europe, US and also China, how do you compare the urban development in China and in the west? It’s a big question actually, you can talk about the most impressive aspects.
最后，作为一个在欧洲、美国和中国都有丰富研究和实践经验的学者，您如何比较中国和西方的城市发展？这是一个很大的问题，您可以就其中某一方面来谈。
S: The way we used to teach planning in Europe in the past was very different from now. Just 20-30 years ago, we still thought there was a best system. Now planning schools around the world teach you to adapt, using words like 'endogenous characteristics' and 'sense of place'. Planners now are much more aware that they need to create more linkages with the local. And even if we are getting similar culturally, planners need to be aware that the location and the regulation make people act differently.
如今在欧洲教授城市规划的方法也和过去非常不同。仅仅二三十年前，我们还认为会存在一个最好的规划体系。但现在，规划院校都在使用“内生特征”、“场所感”这样的语汇。现在的规划师都意识到要创造更多与本地的联系——即便人们在文化上越来越相似，规划师们也必须了解，不同的地域和法规会让人们的行为有所不同。
In the past Chinese people would come here just to learn and implement there, but now they are proud to present their own examples and successful cases. There is also more competition among Chinese universities, striving to produce differently as if in market segmentation.Thus, you have a lot of pioneer projects and you generate your own markets. Now China wants to come here not to copy-paste, but to discuss/dialogue, which I think is what not just academic, but industry dialogue should be.
过去中国规划师们到西方学习，然后回去运用，但现在他们会自豪地展示他们自己的成功案例。中国大学之间的竞争也更加激烈，正在产生类似于市场细分的差异。通过这种类似市场细分的机制，你们有了很多开创性的项目，定义出了自己的市场。现在中国规划师到西方不是要复制粘贴，而是来对话，我觉得学术对话和行业对话正应如此。
L: You have also visited Beijing for several times. You know, Beijing has been paying a great effort to become a true global city. What’s your suggestion for planning in Beijing?
您也去过很多次北京。北京一直以来都在努力建设为一个真正的世界城市。您对北京的规划有何建议？
S: I think it’s clear that Beijing and Shanghai are already global cities. If people had doubts 20 years ago, they shouldn’t have them any longer. But as global cities they should understand the advantages and the risks. Because with an economic crisis like the sub-prime crisis, big cities are the first to suffer, and they suffer the most. The complexity involved inmanaging and sustaining this city on a minute-by minute basis is very hard. So my only advice is: start to include more risk assessment. Make sure that your planners and university students know how to plan for risks, environmental change, and health.
我认为北京和上海已经是世界城市了，这一点毋庸置疑。如果人们20年前对此持保留意见，那现在再也不必疑虑了。但作为世界城市，它们必须理解自己的优势和风险。当次贷危机这样的经济危机来临时，大城市总是最先受到冲击，而且它们受到的冲击最大。要时时刻刻管理好、维系好这座城市，是非常复杂而艰巨的。因此我唯一的建议是：纳入更多风险评估，确保你们的规划者和还在校的学生们知道如何以规划应对风险、环境变化和健康问题。
Countries won’t survive without global cities as we call them the engines of the globe, but I think that in the same way, mid-sized cities and small villages also play an important role in a healthier urban system, and we need to give them good life quality. So even if it’s clear that Shanghai and Beijing are global cities, and they will prosper, never forget mid-size cities and villages, which play a very important role for those major cities.
我们称世界城市为“全球发动机”，它们极其重要，甚至关系到国家的生存，但中小城市和村庄在健康的城市系统中同样很重要，必须确保它们有良好的生活品质。因此，在发展北京和上海的同时也一定不能忘记中小城市和乡村，它们在这些大城市的运转中扮演了非常重要的角色。
本文为《跨学科城市规划研究展望》的精华版。原文发表于《北京规划建设》2016年03期，版权归作者所有
（公众号排版：赵大伟 张祎娴）