For many years insurance companies and fleet managers have used license checking or other lagging indicators to gauge driver risk. While this does correlate to risk, it leaves potential gaps in which high-risk drivers can go undetected. Thanks to artificial intelligence (AI), these gaps are being filled, and insurers and fleets can access the complete risk picture for any driver, at any time.
Plenty of drivers think of time behind the wheel as wasted. It’s a necessary task to get them where they’re going, whether that be work, school, or social activities. In fact, people often choose to live close to work or schools, to reduce time spent on the road. In business, many drivers aim for the fastest way to their destination to spend the least time driving. After all, in business, time is money.
This mindset of spending as little time driving as possible leads to risk-taking. And it’s not only your classic “risk-takers” who take risks during everyday trips. It is well documented that humans take risks based on a built-in risk/reward calculator. In the case of driving, over-confidence in driving ability leads to risk-taking, or at least rule-bending! The American Psychological Association defines overconfidence as “a cognitive bias characterized by an overestimation of one’s actual ability to perform a task successfully, by a belief that one’s performance is better than that of others.” This is one of the reasons many drivers think they are better than other drivers AND they can drive outside of the rules.
As mentioned above, license checks and other lagging indicators are commonly used measures of risk by insurers and fleets. Yes, motor vehicle violations correlate highly to future crash involvement but, statistically, they are rare. This means many high-risk drivers have no violations. Let’s use speeding as an example to demonstrate why. Even if a driver speeds on a regular basis, the chances of getting “caught” are statistically very low.
In fact, in many areas where traffic flow is heavy and average speeds are high, law enforcement will only take notice if the driver is exceeding the speed limit by more than 15mph over the limit. Drivers prone to over confidence quickly learn how far they can push the limit with little or no consequences. This means high-risk behaviors get a tacit reward and, worse, often go undetected with no warning flags for the insurer or fleet.
Telematics was a huge leap forward from traditional lagging indicators. In fact, for many, telematics is considered the “gold standard” in measuring driver behavior. However, to a certain degree, telematics enables drivers to “cheat the system”.
Fleets often find that, when they first introduce telematics, they see a dramatic reduction in collisions and risky behaviors. But it usually doesn’t take long for high-risk drivers to figure out how to operate just below the risk thresholds or discover other ways to “game” the system. Some might argue that telematics therefore does its job – it reduces risk! Yes, but it still allows for a level of high-risk driving behavior. Think of it as an amber zone below the thresholds. Also, let’s not forget that some solutions enable companies to set risk thresholds higher. Companies that choose to do so indicate a willingness to overlook a certain element of risk.
Pattern-based AI takes driver risk analysis to new depths. Using GPS data, obtained via a simple API connection, pattern AI analyzes every part of the driver’s trip. Instead of relying on license checks, previous claims, or even driver events – such as speeding or harsh acceleration/braking/cornering – pattern AI measures every second of driving. In fact, it performs billions of analyzes every hour.
In a similar way to facial recognition, it compares each data point against a database of 7 billion previous trips, each with an associated crash risk and cost. The result is in an individual Crash Probability Score for every driver, making it easy to compare crash risk across a whole portfolio or fleet. More importantly, making it easy to identify high-risk drivers, despite a lack of violations, and who manage to cruise below the telematics thresholds.
High-risk drivers are not only those who speed or display aggressive behaviors. In fleets, slow speed maneuver incidents are common, often occurring while backing up/ reversing. Drivers who don’t speed and who have no license violations could be at high-risk of this type of crash. Yet their driving might not be picked up as risky by traditional telematics solutions.
Similarly, most traditional telematics has no way of contextualizing driver behavior. For example, one driver might hit the brakes hard to avoid hitting a child who runs in the road, while another hits the brakes hard because their attention was taken away due to texting. Telematics would measure the behaviors in the same way. Pattern AI doesn’t, and that’s because it measures continuously, rather than needing to be triggered by an isolated event.
The advantages of being able to accurately identify driver risk level using insights from the whole trip are huge. For insurers, a pattern-based approach to measuring risk overcomes any discrepancies around driving events. It also makes the whole process of insurance pricing fair, and helps insurers optimize profitability. Approximately 15% of drivers cause 50% of crashes. By identifying the 15% riskiest drivers, as well as the 85% who are the lowest risk, insurers can price correctly. The 15% will see their risk level reflected in their premium, and therefore have an incentive to address their risky behaviors to reduce their bill.
For fleets, individual driver risk profiles generated through pattern AI enable early identification of risk and therefore enhance driver risk management efforts. Having a granular understanding of driver risk also enables risk and safety managers to take an educated approach to driver training, ensuring the right drivers receive the right training, at the right time.
Does this mean that telematics data is obsolete? Absolutely not! In fact, pattern AI is another tool in the toolbox for telematics service providers (TSPs) and is even enabling them to monetize their GPS data. Because of advances in technology, TSP customers are looking for value-add from their providers and seeking a one-stop-shop for various needs. Our eGuide on this topic explains how TSPs can use AI to position themselves closer to fleet customers.
With pattern AI, fleets that wish to do so can continue using a black box or other telematics device. This might be for location-tracking, crash detection, or other purpose. Yet with pattern AI, disparate data can be combined into one driver risk score across multiple providers. This gives the end user actionable information to support their business goals and is a reason for TSPs to incorporate AI into their offerings.