The Department of Civil and Environmental Engineering recently launched our new Strategic Directions website, detailing the five principles guiding our vision of the future of pedagogy, research and service. The Building the Future Distinguished Lecture Series provides a forum to discuss each strategic theme and build a broad community that includes industry professionals, researchers, educators, and students. In the final installment, Stanford University Professor of Civil and Environmental Engineering Jack Baker considered ways in which engineers can contribute to the resilience of communities through modeling, flexibility and new tools.
After the lecture, Antoine E. Naaman Collegiate Professor of Civil and Environmental Engineering Sherif El-Tawil moderated a discussion between Baker and Consulting Engineer Chris Poland. The panelists considered the limitations of artificial intelligence, as well as the role of social infrastructure in contributing to resilience.
What follows is an excerpt of their discussion, which has been shortened for clarity. The entire lecture and panel discussion can be found here, along with the rest of the lectures in the series.
Sherif El-Tawil: “How can we incorporate a recognition of the power of social infrastructure into our work on the built environment? Before we go any further I just wanted to define social infrastructure: it is a subset of the physical or built infrastructure that deals with society, mainly schools, congregational spaces, things like that. That’s from our perspective, as engineers to social scientists. [This concerns] relationships: how long have you lived in a community? Is there a crime? Do you really like where you live? If there’s a disaster, would you take that as an opportunity to leave?”
Chris Poland: “You know, the whole notion of the power of social infrastructure from the social scientist perspective is that they really believed that if we took care of the social problems and everybody had a more equal ability for their lifestyle that we would have automatic resilience regardless of the infrastructure. And we know that could be an answer but that’s not this integrated answer. I think the integrated answer is to take a look at at the existing environment that we have — Jack has the models to do that — and to to consider the levels of vulnerability, because the folks who are the most vulnerable are also the ones that are the least resilient, and focus our programs and the design criteria and how we deal with the existing buildings so that we’re really taking care of those vulnerable populations first, the ones that have the most to gain from improving their social infrastructure. And that’s a quick trail to community resilience, if you will.”
El-Tawil: “It used to be that structural engineers focused on the performance of a single building under hazard, but changes in hazard mitigation have shifted the focus to community scale and consideration of many factors. So for the education of engineers, do we need to teach a new set of skills to our students to broaden their knowledge? How do we bring in other disciplines to allow us to look at these city-level problems, to tackle these increasingly complex issues?”
Jack Baker: “You know, to really have an impact, our education needs to kind of follow along with the research. There is a cohort of high-end, practicing engineers that are developing these things and understand them, but to really deploy at scale we need some kind of educational scheme here. At Stanford, two years ago we launched a graduate course in regional seismic risk analysis, where we’re bringing in some of these concepts and trying to formalize that and have projects where the students are kind of working through this type of workflow. And they’re also working on giving presentations to stakeholders, chief resilience officers at cities and things to hear kind of the reactions and hear the pushback when they get too caught up in the numerical metrics and things like that. I think some project-based courses where you’re partnering with stakeholders [are being] increasingly deployed to understand the context specific constraints that a community might have or motivations that they might have, and how that could drive the engineering solutions. As a part of many students’ education there’ll be some new tools in our toolbox. We shouldn’t solve everything. Of course we shouldn’t pretend to be experts in some of the disciplines that we’re not solely responsible for, but I think at least having some appreciation of those disciplines and how our work might interface to the policy world and the economic world and social aspects of community resilience is gonna be really a great asset for the universities to offer. And for those of you students on the call, be thinking about getting some exposure to those ideas. I think that’ll pay dividends in your career moving forward.”
El-Tawil: “Now there was a question during the registration and there’s a similar question in the Q and A, and one of them relates to machine learning and artificial intelligence. The gist of both questions is how can [machine learning and artificial intelligence] play a role in community resilience? I know this is kind of beyond the talk that you gave Professor Baker, but I’m curious about your opinion and Chris, of course.”
Baker: “Yeah, it’s a good question. We want to think about deploying all of the powerful tools that might be at our disposal. I have kind of two, two feelings about it. So one is, even though these cities are tremendously complex and smart cities are producing a tremendous amount of data, in some sense, we kind of have a small data problem. These disasters are fortunately infrequent and are many times unique in their characteristics. Until we have a qualitative understanding of the drivers of resilience, it’s very hard to just feed a small data set to a machine learning model and expect that to replace our modeling. So I think our conceptual research and our qualitative understanding of these processes still needs to evolve and are not going to be replaced by machine learning anytime soon. That’s it; I think there’s a lot of role for remote sensing and smart cities to give us information about exposure or to give us some information about human behavior before and after disaster. To try to understand what are the characteristics of our built environment from imagery and other sources, to understand how people are using infrastructure services by location data, those types of technologies to understand the inputs to our modeling [are] tremendously valuable. I don’t think it’s quite the right application area where just a pure machine learning application is going to solve that problem for us.”
Poland: “It’s hard to understand machine learning and artificial intelligence to start with, but I appreciate that it’s coming. I know that we have areas where we have very, very little information and one of them is just the quality of the inventories, of buildings and infrastructure we have as the basis of our simulations. If there’s some way we can use artificial intelligence to expand the data that we have and learn from, that would be wonderful. But I have to look forward to seeing that develop.”
El-Tawil: “I actually agree with you. I think the very first question that was asked was regarding validation, if we have validated simulation models at the regional level, and if those are multidisciplinary, might they include social science, health and structural engineering? And once you have these validated — what should I call them? Urban scale simulations? I think you could use some of these machine learning models there to kind of run many simulations, teach the model. And when an actual event happens, then you might be in a better position to optimize your response. I think machine learning is very specific. And like you said, Jack, if you have a good training set you’re gonna get a good result for that. The issue right now is that I don’t think we have the data for it, but I think we are going in that direction.
Who do you see as the decision makers? Are they state, county, city, owner, designer? How do these decisions permeate down to the actual design and the legal responsibility of professional engineers?”
Poland: “Who are the decision makers? Well, it depends, you know, one of the best tools that we have to build infrastructure resilience is in the design standards and the building codes, and that rests right with the engineers. There’s a lot of conversation that goes on amongst engineers about what’s appropriate, not appropriate, how we achieve functional recovery levels and all that. From that standpoint, it really rests on the shoulders of the engineers to continue to drive that forward and provide those tools. When you get up to the owner’s level, what they’re really talking about is how they’re going to invest their money, and they make the decisions about whether they’re going to build buildings better or not.That’s really going to be based on what the engineers tell them they’re going to get back from it…. One building at a time is key.”