Using AI to Transition Vehicle Fleets

MBAi and MSAI students developed a model for professional services firm EY to help its clients answer questions about transitioning to all-electric vehicle fleets.

How can a company determine if transitioning its vehicle fleet from gas to electricity makes business and environmental sense?

That was the question professional services firm EY posed to students for the capstone project in Northwestern Engineering's Master of Science in Artificial Intelligence (MSAI) program and Northwestern's MBAi program — a joint-degree program offered between Northwestern's Kellogg School of Management and the McCormick School of Engineering.

Midway through the 10-week course, it wasn't clear if the students would have a concrete answer to deliver. The team worked with anonymized data from the fleet of one of EY's clients, as well as third-party data sources and information.

What became obvious was the vastness of potential solutions.

"What was nice is we got an open sandbox to work in, but we had to be smart on where to go because we had limited time," Brian Schwartz (MBAi '24) said. "Initially we struggled to make those decisions, ended up in analysis paralysis, and did not make clear progress."

It was at that point that Schwartz and his teammates were reminded of an important adage: Don't let perfect be the enemy of good.

“I was surprised by the sheer amount of nuance we could have included in our model,” Hannah Simmons (MSAI '24) said. “Every cost component was like an onion in how complicated it could be. Every component demanded a level of analysis that made the challenge both unexpected and incredibly rewarding.”

That analysis ultimately led the students to a proof of concept tool that provided detailed outputs of how to transition a fleet. The team's model included an analysis of what vehicles to transition, why, and how EY could leverage that information with clients.

"The model that they built is great, it's an optimizer," said David Nichols, a senior principal and global client service partner who leads the EY services team for some of its largest clients. "That's how I look at generative AI, as an optimizer. I thought it was a very interesting case study and they did a great job of applying AI to a very real-world problem."

Nichols, who also is a member of the MBAi Industry Advisory Board, said he wished the students had more time to devote to the project because of the exciting ideas they developed.

Few aspects of their model relied on generative AI techniques. Instead, the students focused more on time series analysis and third party data resources.

“As MSAI students, we are constantly taught to pick the best model for a situation, whether that is AI or not,” Hannah Simmons (MSAI '24) said. “This gave us the confidence to creatively problem-solve without shoehorning familiar but inadequate techniques.”

The capstone project is intentionally designed to mirror realities of the workplace, where technical engineers and business leaders collaborate to solve real problems.

Students from both programs benefited from the experience.

“The MBAi students were crucial in how they always maintained a 10,000-foot view of the project,” Simmons said. “They really helped us get over complex roadblocks and made sure the end product would actually be useful for EY.”

Schwartz agreed.

“While we were thinking through the business problem, the MSAI students were the ones executing and realizing the solution,” Schwartz said. “I learned so much about how to actually realize the optimal solution and know when to pull back rather than always use brute force to move forward.”

The students knew they had a finite amount of time to work on the project, so they made sure to share data sources and various application program interfaces they felt would be valuable for future iterations of the model.

“Discussions have happened within EY’s eMobility group to build a more formal proposal to continue this work into the recommended next phase of model development,” Schwartz said. “Solving a problem that has larger implications when it comes to energy and the environment was very rewarding.”