Generative AI Applications for Fleet Management: Game Changer or Just Another Fad?

Last month, Uber announced the launch of Uber Freight Exchange. This standalone software platform is available to all shippers and carriers. Shippers can run freight auctions with their own carriers, in addition to Uber Freight’s network of carriers. Uber also announced a new generative Artificial Intelligence (AI) chatbot for its shippers. It announced the launch of Insights AI, a generative AI-powered tool that uses Large Language Models (LLMs) to provide insights from Uber Freight’s transportation data. Insights AI currently leverages 45 data points from tens of millions of shipments from the last 2 years, and the plan is to add more. Uber Freight has invested about US$120 million in scaling logistics platforms, data-enabled insights tools, and generative AI since the acquisition of Transplace, a Software-as-a-Service (SaaS) Transportation Management System (TMS), for US$2.25 billion over 2 years ago.

Generative AI has the potential to uniquely enhance the field of fleet management through a data-driven approach with a range of diverse applications that can optimize operations, enhance safety, and reduce costs. Insights from AI-powered algorithms can uncover the hidden patterns of fleet behavior, driver performance, and maintenance requirements that manual methods might overlook. This can result in cost savings through more efficient resource allocation and improved operational productivity. Some avenues through which generative AI can augment fleet operations include:

• Predictive Maintenance: This is an area within fleet management that is in dire need of improvement. Since 2012, the American Transportation Research Institute has reported that maintenance and repair costs represent between 8% and 10% of a fleet’s average marginal cost. By leveraging data points collected from sensors, telematics devices, Electronic Logging Devices (ELDs), and maintenance records, generative AI models can detect patterns and anomalies that suggest equipment or vehicle failures. This allows fleet managers to proactively plan for repairs and maintenance, mitigating risks of breakdowns or unexpected downtime.
• Route Optimization and Dispatch Management: This is another area within fleet management where operational efficiency to ensure marginal returns is key. Generative AI can optimize a fleet’s routes and dispatch processes by accounting for traffic data, weather conditions, and driver behavior. Forbes Magazine reported that U.S. businesses lose US$62 billion per year through poor customer service, up US$20 billion since 2013. Leveraging data to optimize dispatch management and route optimization can help identify areas of improvement and augment route optimization overall.
• Driver Safety and Coaching: Generative AI could potentially analyze driver data to identify risky behaviors and provide personalized coaching to improve safety. Apart from identifying instances of distracted driving and fatigue, generative AI can also simulate scenarios to improve driving behavior.
• Customized Needs: Generative AI can be used to provide customized customer service to large fleets. Deploying chatbots that can instantaneously answer questions and resolve issues can automate monotonous tasks, which can free up fleet managers to focus on more creative and strategic roles.
• Administrative Efficiency: Generative AI can also be used to schedule appointments, generate reports, and answer driver inquiries.

Currently, we are seeing a wave of solution providers offering generative AI-powered applications. Perhaps the most notable one that offers a lot of fleet management applications is Geotab’s Project G. Introduced in a beta launch, Project G offers an intelligent assistant, which is a dynamic chatbot that enables customers to ask questions through a chat interface about their vehicle fleet on critical aspects such as vehicle performance, idling times, fuel economy, vehicle usage, and cost savings, among other things. Project G is linked to the Geotab Data Connector, an advanced analytics tool that extracts and analyzes user data from MyGeotab and turns it into actionable insights. Users can then access these insights through Project G’s easy-to-use chat interface. Fleet managers can directly use the chat interface to draw fleet-related insights—from idling time to driver spending without needing to write Structured Query Language (SQL) queries to generate these insights. Although in its early stages with limited functionalities, Project G is showing promising results. While generative AI offers immense potential for fleets to push toward the next phase of digital transformation, it is crucial to consider certain caveats before rushing to implement.

First, it is important to know that generative AI models are entirely reliant on the quality of data on which it is trained. Biases or inaccuracies in the sourced data will be reflected in the results. Second, generative AI models can also be complicated to comprehend. In its current stage, generative AI should be viewed as a tool to augment human expertise, not replace it. Although fleet managers can leverage insights generated to guide decisions, they should by no means see it as gospel and still deem human judgment and experience as invaluable during decision-making, especially when dealing with complex situations. Continuous monitoring and progress will be imperative to improving generative AI applications in fleet management, and fleet managers should acknowledge this to determine a process for retraining and refining AI models to the fullest extent possible.