This voice is auto-generated. Please let us know if you have any feedback.
As artificial intelligence technology gains prominence, new ways of deploying it can help or hinder progress towards fairer urban transport.
Kofi Nyarko, a professor of electrical and computer engineering at Morgan State University, said AI could make transportation more accessible, affordable and safe. webinar last week Hosted by the Transportation Research Board of the National Academy of Sciences, Engineering, and Medicine. For example, AI can optimize routes for uses such as on-demand transportation serving rural populations.
But Nyarko stressed the importance of “responsible AI” because “AI has the potential to exacerbate or mitigate existing stigma and discrimination in transportation.” Because AI systems learn from data, any bias in the AI training data provided to the algorithm being trained can cause the system to perpetuate existing biases and unfairness, he said. Cities therefore need to ensure that their datasets are diverse, that their communities are fully involved in program design and deployment, and that the system is monitored and improved long after implementation. There is, Nyarko said.
Ziping Wang, Associate Professor of Information Science at MSU, gave an example of her team’s rural drone delivery project. Mr. Wang lives in rural northern Maryland and knows that truck deliveries in rural areas are rare because they are expensive and inefficient.
Mr. Wang described the process of gathering information on local residents’ delivery preferences and shared key oversights that will aid future research. The research team did not believe that the opinions of a particular group (farmers in this case) differed from those of the rest of the broader community. “No prejudice is prejudice,” Wang said. Farmers, unlike other local residents, were skeptical of drone encroachments on their land. Wang explained that the next step in his research is to develop a more comprehensive understanding of delivery demand in rural areas.
Jamie Morgenstern, assistant professor of computer science at the University of Washington, explained how to minimize bias in data from a large and diverse community and ensure data conclusions are reliable. The main point is that data from human sources change over time, she said. “Modeling these changes is very important to make sure the system works as expected,” Morgenstern said. For example, traffic patterns change daily, and neighborhood density may vary by income.
Morgenstern said cities may want to deploy “small-scale studies” like pilots before embarking on large-scale AI projects. She cautioned against making sweeping generalizations based on surveys of program participants, as those who strongly support or oppose the project are most likely to react to it. Morgenstern said hosting small focus groups to improve qualitative data could help alleviate such bias, she added.
