Urban Technology at University of Michigan week 225
Climate Challenges and AI Solutions at Cornell Tech's Urban Tech Summit 2024
👋 Charlie here, making a guest appearance fresh off our program’s latest trip—this time to New York City, where I spent a few days attending the 2024 Urban Tech Summit, hosted by our friends at Cornell Tech.
This year’s gathering was dedicated to the theme of “Intelligence for Climate Adaptation,” exploring how AI and other digital tools can, and already are, contributing to cities’ resilience in the face of climate change. And this year’s trip was extra special, because Bryan and I were joined by fellow Michigan colleagues, including faculty member Matthew Wizinsky and four of our Urban Technology students: Davina Hutapea, Neel Marathe, Dylan Shefman, and Audrey Tang.
💬 Hello! This is the newsletter of the Urban Technology program at University of Michigan, in which we explore the ways that data, connectivity, computation, and automation can be harnessed to nurture and improve urban life. If you’re new here, try this short video of current students describing urban technology in their own words or this 90 second explainer video.
🌦️✨ Climate + AI + NYC
Over two crisp autumn days on Roosevelt Island—sidestepping geese and a one-eyed squirrel en route to the conference venue—the Summit convened speakers from government, nonprofits, startups, utilities, venture capital, academia, and beyond. The first day of the conference laid out the challenges posed to cities like New York by extreme heat, flooding (and drought), and rising sea levels. Day two was a series of closer looks at solutions. Based on this description you might ask if the “Urban Tech Summit” has morphed into a “Climate Tech Summit” after four years, but the reality is that these have always been the same thing. Climate is a motivating purpose of technology in cities, and has been since the first structure was created to provide shade and shelter.
The parts of the event I found most interesting were discussions of projects that exemplified what Rit Aggarwala, NYC’s Chief Climate Officer, described in the opening keynote as being “not about the money… or policy… but in the real problems people face today.” Digital twins are fine, and it’s useful to be able to test different interventions and see their impacts in real-time, but that’s still just a tool. Things like upgrading, retrofitting, and improving utility networks, buildings, and other physical infrastructure are real and pressing needs. Seeing how those needs are met by digital wizardry is where things get interesting.
One of the implicit themes was precision. Tech, and specifically AI, helps people working at the frontline of climate adaptation to understand urban needs with precision so that they can do things like prioritize where best to plant new trees in a city seeking to mitigate heat. Or by generating hyper-local reports on flooding which is the point of FloodNet, a network of sensors across NYC capturing huge amounts of data related to water levels. This empowers communities to develop new tools based on that information.
Precision helps people maintaining the city to spend limited funds more effectively, and it also empowers city residents to make better day-to-day decisions, such as through the possibility of flooding-aware navigation apps as floated by Marquise Stillwell during one of the panels. FloodGen takes that a step further to show how AI can be used to make possible future scenarios palpable. The app displays locations with simulated conditions of mild, moderate, or severe flooding. Looking at Street View of your home is one thing; Street View of your home with 12 inches of water surrounding it is another—both for you and your insurance company.
Here are some of the themes that jumped out to me:
AI and (Human) Capacity
Anthony Townsend highlighted this in his keynote address. There’s both a supply question here—we won’t have the number of engineers and scientists we think we need for climate action, so how do we make up that gap?—and a related question: if AI is able to do some types of work, what does that free up humans to tackle?
One refrain throughout the conference from people working in government was that there is often too much data available, so making sense of it is the first problem to be solved. But that’s actually an area where AI is already quite useful. Anthony cited a project where students at Cornell Tech collaborated with Accenture to gather and then synthesize over 200 resilience plans that have been developed by cities around the world (here are New York’s and Singapore’s, for example). Rather than someone spending the time to read and analyze those plans one by one, artificial intelligence can rapidly highlight the number of occurrences and weight of common themes and approaches across cities, in a fraction of the time.
So if AI can help with tasks like this, what does that leave humans to do? (What will our urban technology students need to do?) Sitting there I was thinking back to Anthony and Bryan’s past conversation about future urban technology jobs, which is a topic we will need to continue refining as technology advances.
AI and Precision
Bracketing the question of what we mean when we say “AI” (a few speakers noted that we should be more precise talking about the differences between language models, machine learning, computer vision, etc.), another common theme I heard was that what will be more useful for climate challenges are AI models that are focused, lean, and specific—built to deal with a given data set and for a particular purpose, rather than trying to use massive, generic models for all projects. In addition to yielding better results, that approach will be faster to train and less resource-intensive to operate (which makes them more cost-effective too).
Speed and Scale
Climate challenges are large and rapidly growing, so there was an urgency to all of the conversations about climate solutions at the summit; and speakers also underlined how climate solutions needed to make financial sense and be scalable in order to have a significant impact.
That urgency, unfortunately, often runs into a wall when it comes into contact with government processes and procedures, especially around procurement. As one panelist said, the government is good at building complex things that need to last a long time, but it’s less good at building big things quickly or prototyping and then rapidly iterating on an approach.
One of the surprising points of agreement at this technology summit was that government procurement methods should be revisited. Not the high-tech topic I was expecting to generate excitement! A promising approach is challenge-based procurement, recommended by Cornell Tech as part of its Pilot: New York City program, where rather than a traditional, highly-prescriptive and contract-based government procurement process, agencies are able to describe the problem they are trying to solve and solicit solutions from the private sector.
Once possible solutions have been identified, they are piloted and tested prior to a large, formal purchase. This helps cities get their needs addressed with innovative solutions and also helps build a market for startups (hopefully local ones) building those solutions. Stacey Matlen (a Michigan alum!) spoke about her work with the Partnership Fund for New York City and their Innovation Labs, which are focused on specific themes like transit, buildings, and the environment. In each case they’re sourcing and pairing private-sector solutions with public-sector agencies and problems.
After two days, I walked away from the Summit and onto a delayed flight (due to weather, fittingly) thinking about the future of climate technology and wondering where our Michigan students will make a difference. As dizzying as the different tools and applications of AI might be, I was heartened to see them applied to real problems. For our students this provided many good examples of what you can do with a career in climate/urban technology. Scores of people working on climate adaptation in different ways were united by their shared goal of keeping cities stable and livable despite climate change, even as the tools and approaches they discussed were as divergent as AI, procurement, finance, data visualization, and fly brains (seriously). This highlighted an important and more fundamental kind of precision: the precision of having a mission centered on real problems, experienced by everyday people, rooted in the specificity of place.
These weeks: Gearing up for final reviews, registration for next semester, pondering what turkey technology might look like. 🏃