An urban mobility network that harmonizes mass transit, shared mobility, and walking and cycling is the goal of many city government plans. Yet the true potential of shared mobility is not yet realized, says David Shmoys, a professor at Cornell University’s Center for Data Science for Enterprise and Society and a contributor to Citi Bike’s ride-sharing algorithms.
Many hailed rides are not shared by multiple passengers, and the service is not properly used in harmony with other urban mobility modes, Shmoys explains. But artificial intelligence (AI) and data tools could give mobility providers the nuanced and more granular impressions of their customers they need.
“This is an AI and data science problem,” Shmoys says. “We don't have the tools yet to understand who the customer base is to be able to market the right kinds of shared mobility products in a way that they get used effectively, and then to layer that with the actual logistics of making the shared experience.”
Shmoys recently spoke with Tom Fleming, a partner in Oliver Wyman’s Energy and Natural Resources practice, about AI’s potential in mobility.
How will the mobility industry move forward with AI?
The largest impact currently has been the speed that AI coding processes can build a platform to serve a given customer base. Catching up to, for example, ride-hail incumbents is extraordinarily enhanced by AI.
What I think has a much larger promise is the ability of AI tools to allow real-time response and forecast mobility demand for ride-share companies.
More specifically, how can AI change the ride-hailing industry?
Ride-sharing is not where people expected it to be currently. The word “share” is in the term, but the amount of vehicular sharing that occurs is dismal from my perspective. It would be tempting to abandon the idea altogether if the environmental promise of it wasn’t so great.
We've supplanted one taxi system with another, and the key is to get people to share vehicles with the right market group, pricing, and effectiveness. If ride-share providers put consumers in cars that for whatever reason they feel more comfortable with, then it’ll change the user experience. And with a higher occupancy rate in vehicles, then there will be a much more efficient system.
What are your thoughts on use cases for CEOs who want to use AI?
One element of AI’s potential is what I call the automation of the logistics pipeline. We now have the data for a firm to say, “If I had known the outcome of a policy in advance, would it have been the right policy?”
AI can help firms build retrospective models that can be revolutionary for building new use cases. These models can empower firms to reconsider the counterfactuals against a variety of other policies they may be thinking about and build probabilistic models of what the future looks like. That will give a more resilient set of tests to plan against the future. As one analyzes the performance of large-scale mobility systems, I think AI hasn’t been used to the extent it could be.
What are your thoughts on the multimodal network for cities? Is it successful?
I think it has failed thus far because there are no tools that have truly honed public perception to convince travelers to take an e-bike, to take the subway from point A to point B but then hail a ride to take them the remaining journey. There’s enormous potential for a tool like this, but users are too specific in how they’re willing to behave and their degrees of flexibility. Mobility providers need a more personalized, probabilistic model of their clients.
The same potential is similarly lost in other mobility sectors due to a perception problem. The promise of commercial electric and autonomous vehicles 10 years ago, for example, was that they would have an easier uptake potential than passenger cars – particularly for trucking. And yet the opposite happened: autonomous and electric passenger vehicles both captured the imagination of the public and as such the ability to drive investment is more limited for commercial applications.
Sometimes consumers need several mobility apps to pay, plan, and find available modes. That may harm public perception of multimodality’s potential. Who is the best positioned to be the connector between all the options for shared modes and consumers?
I think it’s up for grabs.
I wonder about modeling other cities like Shanghai, where centralized government resources could build such solutions. That's a diametrically opposed mechanism for which you have the sort of information sharing and the capital investment to build something.
One could imagine a medium-sized shared mobility player deciding that this is going to be a priority and thinking about their service as tailored to being something that works in conjunction with mass transit, micromobility, and good, old fashioned walking. It’ll require a large investment – perhaps from a platform like Google that realizes they have the capital and the partnerships with other players to put it all together.