AI In Mobility: Successes And Barriers In AI Adoption

OliverWyman Forum
key takeaways

Thank you for joining our discussion on AI and mobility. The industry is excited about the technology’s potential to dramatically improve transportation services and user experiences, but companies have significant technical, financial, and behavioral hurdles to overcome. Our three expert speakers – Fabien Cros, data and AI lead for manufacturing at Google Cloud France; Lennaert de Boer, director of technical program management and product operations at Cruise, and David Shmoys, professor of business management and leadership at Cornell University – provided sharp insights about the road ahead. Here are a few of our key takeaways:

 

  • Mobility companies need to develop a data mindset. The culture of the auto industry is rooted in mechanical engineering and keeping tight control of branding and design. That strong internal orientation impedes the close collaboration – both within the industry and with technology providers – that’s required to develop AI applications and drive mass adoption among consumers. “Traditional OEMs are used to supply chains, building cars, buying spare parts,” said Cros. “They are not used to going to clients and suppliers and saying, ‘How can we exchange data?’ You need to change the entire organization toward data monetization and data as a product.” Academia has a role to play here. Shmoys said the old engineering framework, which categorized people as either working with bits or with atoms, is outdated. “The cyber-physical divide is no more,” he said. “We need to educate engineers with an AI ready and data centric point of view.”
  • Good data processing takes a lot of time and money. AI is only as good as the information that gets fed into the system, and getting that right isn’t quick or easy. “Tagging, annotating, and cleaning data are not sexy, but they are the foundation of AI solutions,” said Cros. “You can have the best LLM ever, but if you don’t have this underlying foundation, there is not much you can do.” In addition, the explosive growth of generative AI and increasing complexity of models is increasing prices for everything from semiconductors to the AI models themselves. “It’s driving up prices significantly and making the feasibility of offering a low-cost generalized tech solution very low, which makes it very hard for adoption to take place,” said de Boer.
  • People are as important as hardware and software. Companies need a deep understanding of consumers’ anxieties and desires to drive adoption of AI in mobility. While there are plenty of eager, early adopters of the latest technologies, many others may be unfamiliar with the new and more comfortable with the old ways of doing things – preferring a taxi to an Uber, say. Companies need to acknowledge this and explain to consumers how AI can make their lives simpler, easier, even safer. As Shmoys put it, “Technology is adopted when users are incentivized by their own convenience in a significant way.” The average person today may be less tolerant of accidents caused by autonomous vehicles (AVs) than by human drivers, but eventually the data may help win over hearts and minds if it continues to show a large safety advantage for AVs.
  • Generative AI has a role to play in winning that safety argument. Most people find driving becomes second nature after a while, but getting from point A to point B on crowded city streets or packed superhighways is an incredibly complex process. No matter how many miles we drive, there are always new scenarios that can arise – erratic driver behavior, poor visibility, a deterioration in the roadway – that demand an immediate reaction. There simply isn’t enough real-world data an AI model can draw on to respond to a 0.001% risk situation. “You’re going to need synthetic data to model those scenarios,” said de Boer, “and that’s where gen AI can be very helpful.”
  • Get used to uncertainty being the new normal. It takes years to design and bring a new car to market, and a decade or more for a commercial aircraft. By contrast, technological innovation advances at warp speed. “You have no idea what the car of tomorrow should or will look like,” said Cros. “You have no idea what the user will want. For manufacturers, it’s a nightmare.” Mobility companies need to build agility into their product development and manufacturing operations and keep their fingers on the pulse of the market through constant feedback from consumers.

We want to thank everyone who attended our virtual event and contributed to the discussion. We all have a big stake in the ways that AI is implemented in the mobility sector, whether as members of the industry or as consumers and citizens. We look forward to continuing these conversations and to sharing with you our research into AI and mobility when we publish our next report, “What Businesses Can Learn From Mobility’s Early AI Adoption,” in September.

Event Overview

AI is often portrayed as a safety risk, but the mobility industry has a long history of using the technology to make travel safer, more efficient, and better for consumers. Its experience should inspire decision-makers in mobility and beyond as they brace for transformation.
 
AI in Mobility, an upcoming report from the Oliver Wyman Forum, combines interviews with experts from industry, academia, and the public sector with global consumer survey data on AI, to provide a unique analysis of the mobility industry's approach to AI and what other industries can learn from it. 

 

At our virtual launch event, the report's author, Andreas Nienhaus, explained the findings. A panel discussion with leaders in industry and academia followed.

 

Event Highlights:

  • Introduction: Introduction from Andreas Nienhaus on the research.
  • Panel Session: Leaders in industry and academia will share their perspectives.
  • Q&A: Attendees will have the opportunity to ask questions and engage with experts.