Discovery Insure Study Finds AI Coaching Cuts Risky Driving

In Short

• Discovery Insure says personalised AI coaching can reduce risky driving behaviours.
• The study covered 5,000 drivers, 1.7 million trips, and 15 million kilometres of driving data.
• The biggest improvement was a 9% drop in cellphone use while driving.

What Happened

Discovery Insure has published results from a new study on personalised AI coaching for drivers. The research was done with Johns Hopkins University and behavioural consultancy Affective Mobility.

The study tested “hyper-personalised” coaching. That means the guidance changes for each driver and is delivered at the moment it is most relevant, for example after harsh braking or during a speeding pattern. The coaching is AI-powered, meaning software analyses driving data and decides what message to send and when.

Researchers used real-world telematics data. Telematics is sensor-based driving tracking, similar to how a smartphone or in-car device can record speed, acceleration, braking, and phone usage.

Compared with a matched group that did not get the intervention, drivers who received personalised coaching improved across five risky behaviours:

• Cellphone use while driving, down 9%
• Harsh acceleration, down 5%
• Harsh braking, down 5%
• Harsh cornering, down 4%
• Speeding, down 3%

Discovery Insure also pointed to its longer-running Vitality Drive programme. It says its most engaged users have much lower fatality and accident rates than the national average.

Why It Matters

Insurance and mobility firms in Africa are collecting more driving data, but turning that data into safer roads is harder than scoring drivers after the fact.

This study suggests AI-driven coaching can shift habits, not just measure them. Small percentage changes may sound minor, but driving behaviours repeat across thousands of kilometres, so the risk reduction can add up.

For insurers, fewer accidents can mean lower claims costs and better pricing models. For drivers and fleet operators, this approach could also support safer driving without relying only on penalties, it can combine feedback with incentives.

The results also highlight a clear target, phone distraction. A 9% reduction in cellphone use is meaningful in markets where distracted driving is a major contributor to crashes.