AI & The City
AI is the technology that will have the single biggest impact on cities over the next decade. At Urban.Us, we already meet teams using very large datasets to train algorithms to drive cars, water yards efficiently, guide drones to survey construction sites and route first responders to the people who need them most, and others who use bots to provide legal guidance to people with parking fines.
These startups are benefiting from an explosion of data generated by human activities and sensors. Ironically, while the flood of data is difficult for people to understand, it’s great for teaching machines. Thanks to cheaper storage and processing to train new algorithms, we’ve seen a surge in AI deep-learning techniques. Most people have experience these through performance jumps in areas like voice services like Siri, Google and Alexa. For more on why AI applications are surging check out this backgrounder from The Economist.
AI Will Transform Cities
AI is great for a number of city challenges, but it creates new social, policy and legal issues. In September 2016, the AI100* study panel shared their report on AI and life. The goal of the report was to educate the public as well as businesses and government to help them understand AI and better plan for its governance.
The report focused on a typical North American city to highlight specific changes that will impact people every day. These are the eight areas where the AI100 panel believes AI will have the most impact:
+ transportation
+ health care
+ education
+ low-resource communities
+ public safety and security
+ employment and workplaces
+ home/service robots
+ entertainment
We’ve added emphasis to highlight the areas in which we have chosen to seek out startups to improve city life. We’ll briefly explore 3 areas of impact: transport, public safety and economic opportunity.
Transportation
Autonomous transportation will soon be commonplace and it has some very specific risks. Autonomous cars will be many people’s first experience with physically embodied AI systems. As a result, these interactions might define levels of trust and broader reactions to other types of AI.
It is, therefore, appropriate that the US DOT recently released a 116-page policy document for automated vehicles [PDF]. The report focused clearly on the public benefits, highlighting the current safety challenges and making the case for tech to be deployed to make vehicles safer. However, the report is also very clear that autonomous vehicles mean that the DOT will need new tools, authorities and regulatory structures to ensure that the public interest is served.
Public Safety
Cities have already begun to deploy AI technologies for public safety and security. By 2030, the typical North American city will rely heavily upon them. These include surveillance cameras that can detect anomalies pointing to a possible crime, drones and predictive policing applications.
In her recent book, Weapons of Math Destruction, Cathy O’Neil offers examples of how algorithmic approaches can fail in the absence of clear feedback loops that help them to make corrections. For law enforcement, there is a real risk that the algorithms will automate and institutionalize some of the biases that are already a part of some interactions between the police and different members of society.
Economic Opportunities
AI will likely replace tasks rather than jobs in the near term, but if history is any guide, there will be new jobs that we cannot yet predict. However, it is easier to map out what existing jobs may be lost. At the heart of this problem is migration: people are moving to cities in search of economic opportunities and jobs are most often at the top of a mayor’s agenda.
So how quickly will jobs be destroyed versus those created in the face of AI? Many people working in technology have made the connection between AI and basic income, because they believe AI is will profoundly restructure the labor market and force us to find a way to compensate people in ways that are not related to labor. The Economist explores the possible relationships between AI and jobs.
AI Policies
All of these changes raise new questions for regulators and policymakers. Who is responsible when a self-driven car crashes or a public safety algorithm sends first responders to the wrong building? How can AI applications be prevented from scaling biases like racial discrimination? Who should reap the benefits of efficiencies enabled by AI technologies and what protections should be afforded to people whose skills are rendered obsolete?
AI implies imperfection. Though AI is going to outperform humans in a growing number of tasks, failures are inevitable. Beyond the policy and regulatory issues, one of AI’s great challenges may be trust. When people die in car accidents, we will take little comfort in knowing that statistics still show that fatalities occur less often with an AI driver.
AI is the most important technology ever used to design and manager human settlements. It is going to enable unparalleled improvements. Unlike past technologies, like steam power or the automobile or the internet, many of the gains will be made without humans, even when they are made in the name of humans. Now we need to help regulators and policymakers to understand quickly some of the coming changes, so we can maximize the benefits to the world’s cities.
*Peter Stone, Rodney Brooks, Erik Brynjolfsson, Ryan Calo, Oren Etzioni, Greg Hager, Julia Hirschberg, Shivaram Kalyanakrishnan, Ece Kamar, Sarit Kraus, Kevin Leyton-Brown, David Parkes, William Press, AnnaLee Saxenian, Julie Shah, Milind Tambe, and Astro Teller. “Artificial Intelligence and Life in 2030.” One Hundred Year Study on Artificial Intelligence: Report of the 2015–2016 Study Panel, Stanford University, Stanford, CA, September 2016. Doc: http://ai100.stanford.edu/2016-report. Accessed: September 6, 2016.
