Fleet management teams will be augmented by “always-on” AI agents dedicated to performing rote tasks, allowing managers to focus on higher objectives.
The concept of AI, or artificial intelligence, seized the zeitgeist with the introduction of ChatGPT on November 30, 2022, and has only grown since. Yet fleets have been using versions of AI for some time.
AI tools are used today to predict parts failures, optimize routes, and manage fuel.
Telematics systems use AI to monitor safety alerts in real time and analyze driver behaviors to predict the likelihood of crashes. Video telematics systems using cameras employ AI to identify, organize, and score events such as internal cab distractions or external conditions such as dangerous vehicle maneuvers or hazardous road conditions.
Using machine learning algorithms, these “predictive AI” systems spot patterns and predict outcomes based on historical data.
At Element Fleet Management, “We have a variety of examples using AI that assist fleet managers,” said Element’s Vice President of Strategic Advisory and Client Analytics Steven Jastrow.
When analyzing tire wear, Element uses machine learning to identify replacements that were caused by getting a flat tire after running over a nail. “We can eliminate that replacement when trying to determine how many miles a tire will last,” Jastrow said.
Small AI Applications Are Growing
These small applications are growing. Fleet management companies, telematics providers, and fleet service platforms are in various stages of implementing AI in their toolsets and offerings.
One benefit is total cost of ownership (TCO) modeling.
“While (TCO) can be calculated manually today, we plan to compute it at scale and layer predictive analytics to create optimal replacement timing,” said Chief Client Officer for Mike Albert Fleet Solutions Chris Parrott.
The next step is prescriptive AI, which crunches collected data to help users make even more informed decisions on replacement timing, route planning, and maintenance to minimize downtime.
“Predictive is great but what matters is how easy it is to take advantage of these insights,” said Shiva Bhardwaj, founder and CEO of Pitstop, a predictive maintenance software provider. “Prescriptive is the ability to take the insight and add as much detail that guides the action to be taken.”
Though these tools are becoming more refined, they don’t foretell the true power of AI.
“This type of statistical modeling is not AI in the manner most people think about AI,” said AI Strategist for Merchants Fleet Dr. Suzannah Hicks. “But it does fall into the AI category as machine learning in doing something a human can do, which is the basic definition of AI — a system that can perform a task that a human can perform. When people think about AI, people typically think about self-driving cars and what we are now seeing with Generative AI.”
Next Steps: Incorporating GenAI
Enter generative AI or GenAI, which is the foundation of ChatGPT and competitors like Google Gemini, Claude, Perplexity, and Microsoft Copilot.
GenAI can create new content such as text, images, videos, audio, or even software code in response to user prompts as part of a question-and-response dialogue.
In other industries, ChatGPT is being used to analyze medical and lab records to provide real-time insights into patients’ health conditions. In education, generative AI grades assignments and creates lesson plans. In the legal profession, ChatGPT can analyze contracts to ensure that they are legally binding while providing insights into case law and legal precedents.
“There are new use cases for GenAI being discovered daily,” Hicks said.
In insurance, pricing models and claims cost assessment tools require the accumulation of historical statistical information and the assessment of real-time dynamic market costs.
“Being able to assess two streams of data simultaneously gives commercial auto insurance professionals a streamlined source for better risk selection and improved loss ratios,” said Element’s VP of Insurance Angelique Magi.
Element is also investigating ways to enable a more Google-like user experience to provide responses to common questions based on large amounts of text and analytical data.
According to Element’s Director of Digital Solutions Samantha Rosati, support agents can use AI to digest a variety of documents and then answer policy questions or provide specific guidance. An analyst can also ask data questions and return responses with validation rather than running various reports and performing the needed analysis.
GenAI in Maintenance & Fleet Workflows
Pitstop is using GenAI to simplify vehicle fault codes. When a fleet’s internal database lacks a description for a specific code, ChatGPT provides the missing information and definition, Bhardwaj said.
Compared to other forms of AI, “We’ve found the most value in generative AI,” said Drake Bauer, CEO and co-founder of Flete, a workflow management tool for fleets.
For repair shops, AI can revolutionize daily planning by factoring in backlog, job length, and other priorities to accurately predict how long each job will take. Instead of relying on general estimates, fleet operators will know precisely when their vehicles will be back on the road.
Bauer said Flete is using GenAI to digitize dispersed documents and data, such as paper receipts stored in file cabinets, sticky notes on desks, and whiteboards in the maintenance shop. Taken further, those records will be automatically normalized with other records such as tolling data.
“Many folks we talk to are still committing hours every week to log data into multiple tools and spreadsheets, booking service appts, and creating PivotTables for leadership,” Bauer said.
“Instead, leveraging AI to automate these workflows will give fleet managers more time to spend on tasks a computer can’t perform.”
AI: Fleet Managers Weigh In
Fleet managers have been using AI’s more traditional capabilities through their video telematics systems to optimize routes and prevent distracted driving. GenAI is new territory, though some have been “dabbling in it.”
Ted Chan of Schindler Elevator has been aggregating policies using AI and comparing them to internal policies to help refine best practices. He said it’s been a time saver from having to collect those policies manually.
Thurman Register of Ferguson, a construction services provider, has been using GenAI to produce ad hoc reports and answer specific driver questions that he or a fleet staff member would have answered manually.
But mostly, fleet managers are looking forward to employing AI in new use cases.
For Register, a longer-range objective is to apply AI to maintenance, fuel, and inventory programs.
Chan is interested in the potential for AI’s predictive output to forecast vehicle buys and leverage better decision-making around ownership costs.
Billy Wassmer of TruGreen is looking to use AI to augment his use of Power BI to make better data-driven decisions.
“I believe there is so much more we can do with the data coming from our fleet,” he said. “As costs continue to rise and as safety is always paramount, data will help us be better in both spaces.”
How Smart Can AI Get?
AI users can look forward to their own AI assistants. Earlier in 2024 Geotab released Geotab Ace, a GenAI “copilot” designed to provide fleets with data insights.
“An always-on AI agent will continuously look for changes in the data or video and rapidly recommend corrective action, such as safety coaching or vehicle repairs,” said Geotab’s VP of Data & Analytics Mike Branch.
Branch said that when autonomous transportation arrives, AI will predict maintenance needs, schedule repairs, and arrange alternate driverless rides.
“We are merely at the precipice of realizing AI’s potential gains,” he said. “Currently, it is widely being used as a productivity boost via chat interfaces, which is only a fragment of the value it can bring to the world once access to AI agents is democratized.”
The next steps might sound more like science fiction, but the groundwork is being laid to bring AI from machine learning to the foundations of machine consciousness.
Jastrow illustrated the point that while medical websites can diagnose health symptoms today, the evolution of AI will include asking follow-up questions to better diagnose an ailment. “Today AI will only do what a human has programmed; in the future, AI will develop its own solutions,” he said.
“Right now, AI is the least capable we will experience it,” Hicks said. “AI, including GenAI, will continue to get smarter and generally more capable of completing tasks that humans currently do.”
AI: Proceed with Caution
Along with GenAI’s great promise, the potential for misuse stems in part from AI’s immediate accessibility and our ability to implement it at extraordinary speed and scale.
Pitstop’s Bhardwaj’s breaks down areas of concern:
- Data privacy: Protecting sensitive fleet data from unauthorized access.
- Algorithmic bias: Ensuring AI models are fair and unbiased.
- Job displacement: Managing the transition to an AI-driven workforce.
- Cybersecurity: Safeguarding fleet systems from cyberattacks.
“Responsible AI should also be grounded in ethical considerations, fairness, and transparency,” Branch said. “Businesses considering the possible effects of replacing functions with AI-powered technologies should investigate strategies to retrain and upskill their staff.”
Branch also raised the need for enhanced sustainability practices, as AI’s computing power demands will produce substantially more carbon emissions.
Bauer cautioned that an over-reliance on automation and a “cost-cutting first approach” risks neglecting human oversight in critical fleet operations.
Jastrow is also concerned about the loss of skills as AI does increasingly more work. “Leaving tasks to machines is good; however, as humans, we still need to identify problems, both potential and actual, and then be able to pivot along the way,” he said.
Another specific problem is “hallucinations,” where GenAI produces content that is not based on factual data or reality. “At the end of the day, humans are responsible for the content, which is why we will always need humans, among many other reasons,” Hicks said.
Enabling, not Hindering Fleet Jobs
The interviewees for this article don’t necessarily believe AI will eliminate fleet jobs; instead, they believe it will enable managers to focus on higher objectives.
“We believe the number of vehicles under a single fleet manager will grow significantly,” Bauer said. “Automating every action from the beginning to the end of an asset’s lifecycle might be configured in just a few clicks.”
Yet there are dangers for those who don’t engage the technology. “Fleet professionals should embrace AI or risk relinquishing a competitive edge,” said Parrott of Mike Albert.
Harnessed properly and with the right controls, AI is poised to be an immense benefit to fleet management.
“I don’t view AI as a threat to my job, but rather as an opportunity for enhancement,” said Mikhaila Baldwin, fleet manager for Coolsys. “AI has the potential to generate innovative solutions that might not have been previously considered.”
“I think many fleet managers are under-resourced,” said Chan of Schindler Elevator. “AI can only assist me in my role.”
“I think it will be more and more our co-pilot.”
