The rise of artificial intelligence (AI) was destined to happen, but we didn’t see it entering our everyday lives as it has. Over the past decade, we’ve seen more AI-integrated devices take over the tech industry than ever before. It’s now becoming a tool for writing documents, tracking a home’s energy consumption, and compiling statistical data through algorithms created by software developers.
The AI boom coincides with the other major technological revolution: cleantech. Renewables are scaling up at a rapid pace, and more tech-savvy distribution systems are using AI to track energy consumption and emissions.
Electric vehicles (EVs) are no different. Many cars use AI in the infotainment system, like Rivian, which uses Amazon Alexa in its cabin. EV charging stations are using some form of AI tracking, but what does that mean for drivers?
Research from the University of Michigan and Ulitidata suggests that AI-powered EV chargers offer insights we would never normally see. AI can spot mechanical problems, find out why some are inconsistent with their power draw, discover why equipment wears out, and explain how chargers impact the electrical grid.
Photo Courtesy Oxana Melis
The Business Download spoke with Bob Rogers, PhD to discuss the implications of this new tech. Rogers is a Harvard data scientist and the CEO of Oii.ai, a software modeling company that improves supply chains through AI solutions. He served as Intel’s chief data scientist for three years, and in 1993, he published a book about AI’s future potential. As an EV owner, he understands the technology required to charge cars and how many charging stations are remote data centers for some EV companies.
“The biggest challenge with an EV, especially when you need to charge outside the home, is there can be huge variations in the backlog,” Rogers, a Nissan LEAF owner speaking from his home in Hawaii, explained.
“Living in Hawaii, folks don’t have the capability to charge, so we’re all battling it out here on the grid.”
He believes the motivation for EV charging providers is using AI to optimize revenue and access to stations. AI can be used to reward customers if there is a “network lock on prices” where consumers either pay more or less than their peers. Rogers hasn’t seen this policy play out yet.
“If I’m paying more, and I think the strategy they’re wanting employees to have folks to have preferred customer programs and things where you have a little bit of network lock to maybe give lower prices on average to those folks,” Rogers continued. “Then balance it out with the higher prices for others. As a consumer, I would say so far I haven’t seen any good strategies work where you’ve got a good way to spread out the congestion, but that would be very desirable.”
Rogers explained that Oii.ai doesn’t help optimize charging systems; however, it can create supply chain models for EV businesses. He notes a paper he read about two AI components where one uses reinforcement learning models to “learn what strategies you should use to offer the service and incentivize people to spread out their demand.”
“The connection to what we do at Oii is that we use simulation to figure out how to smooth out the congestion in a distribution network, not just products on the road, but products in warehouses and products moving to a factory,” Rogers explained.
Oii can simulate how much production space a company needs and how often it needs to optimize growth in distribution and sales. Rogers could create a simulation for all facets of the EV supply chain, from critical mineral sourcing to a car’s sale. That would include the charging components as well.
Photo Courtesy Sophie Jonas
“The supply chain you’re describing is an interesting variation where you have one or two very complex products based on many sub-components,” Rogers said. “There are different kinds of optimizations.”
“It’s almost like threefold because you have to not only do the battery, but you have to do the whole car itself,” he continued. “Then you have to do the charging, and there are charging components that make it a crazy supply chain. I’m sure having a simulation like that would help a lot of people.”
AI is certainly exciting. It offers several benefits for EV owners looking to save money on charging. However, it is not all fun and games. There are multiple ethical dilemmas plaguing AI that are not exclusive to ChatGPT or image generation. EV charging faces problems like data stealing and privacy. Rogers said these are some of the main concerns for EV customers.
“The core benefit is for whatever set of resources we have, in the case of EV charging, it’s the different charging stations and the available electrical capacity and the number of, you know, the the number of people who have demand you’re really able to get the most productivity out of that resource, right?” he said. “When you can also operate that resource efficiently. It means it’s a viable business model to make more of that resource.”
Data privacy in EV charging remains a question that many automakers and legislators haven’t been able to answer.
It’s becoming increasingly hard to navigate the data mining area as tech companies and car manufacturers enter into business with each other.
One concern cited by Ford when it joined the North American Charging Standard (NACS), the Telsa charging standard most automakers are adopting this summer, is the possibility of Ford data being tracked and stolen by Tesla and vice versa. Ford is worried Tesla will steal its code to make improvements. While it’s widely believed there wouldn’t be a benefit to stealing another company’s data, you never know what advantages could be found.
Photo Courtesy Markus Spiske
Of course, EV chargers stealing car data would open the conversation about the ethics of data tracking, which could cause a drop in consumer interest in EVs. Rogers believes this is one of the main problems of AI.
“The big challenges are things like, ‘Well, what data is needed to do that?’ Does somebody have that data, and are they infringing on my privacy?” he said.
Rogers said another is the collection algorithm and whether those are safe for consumers. It needs to be trained to be private and protect individuals.
Rogers is also the founder of BeeKeeper AI, a secure technical training platform for federated data. He has used his software in areas like healthcare to fight data bias. He helped create the first U.S. Food and Drug Administration-cleared AI X-ray device at the UCSF Medical Center. Federated data doesn’t ever change, allowing companies to develop an algorithm without impeding on people’s privacy.
“If an algorithm is testing my price sensitivity, or using a charging station or a group of charging stations, and it could be testing my price sensitivity to any product, they’re starting to learn how to manipulate me in terms of ‘Yeah, it’s available, I’ll pay whatever price.’ Am I pushing back? Am I wanting other information? That, along with sensitive information about what I do every day, where do I go? How much time do I spend that a well-crafted algorithm could compromise my source of autonomy by finding the knobs to control me and my buying behavior?” Rogers explained in an example with EV charging and BeeKeeper’s capabilities.
He still believes people should be concerned about their data, though. The hope is that privacy-securing technologies come out to stop data privacy violations.
EV charging is getting more convenient as more stations open across the country. AI integration can make some of the user experience more efficient. However, we don’t know what AI is fully capable of, nor do we know if companies are stealing data from EVs upon plug-in. Even if they are using it to make informed decisions about car performance, more questions and concerns will indeed be raised as the country continues down the road of electrification.
