A joint White Paper by insurance provider Discovery Insure, US research university Johns Hopkins University and UK-based behavioural consultancy Affective Mobility posits that drivers who receive hyper-personalised, AI-powered, in-car coaching record measurable improvements across five risky driving behaviours.

The study: “Personalisation: The next frontier in behavioural change. How habits and AI-driven personalisation are reshaping driving risk”, analysed 5 000 Discovery Insure clients across 1.7-million trips and 15-million kilometres of real-world driving.

Road accidents are usually not random but are attributed to driving behaviours that are repeated and become the habits that steadily increase accident risk over time.

The research has shown that a new approach to changing these behaviours as they occur: hyper-personalised, AI-powered coaching delivered at the point in each trip when it has the greatest impact, steadily reduces accident risk over time.

Relative to a matched comparison group, drivers receiving personalised coaching achieved measurable improvements across five risky driving behaviours, with the largest improvement seen in 9% improvement in cellphone use while driving, with this highlighted as a major cause of accidents.

Improvements were recorded across other problematic behaviours, including 5% in harsh acceleration, 5% in harsh braking, 4% in harsh cornering and 3% in speeding.

Discovery Insure CEO Robert Attwell pointed out that Discovery Insure has, over time, built a base of driving data spanning 22-billion kilometres across 31.5-million client months.

“Against the backdrop of this rich dataset and the increasing availability of AI, we asked: could the individual behaviours behind risky driving habits be changed through deeper personalisation – actively preventing risk well ahead of harm occurring?”

Therefore, Discovery Insure’s data capability was matched with Johns Hopkins University’s academic expertise, supported by the insight-motivation-action (IMA) framework developed by behavioural scientist Dr Neale Kinnear, who is the founder of Affective Mobility.

“The IMA framework helped us design coaching around what drivers need: insight into their own driving, motivation to improve and practical steps they can take. It gave the study a clear foundation in behavioural science and helped us test how AI-enabled personalised coaching could build on Discovery Insure’s already successful telematics programme,” he says.

The 5 000 participating drivers were randomly assigned across experimental groups to compare four intervention strategies – coaching style (directive versus supportive), goal design (behaviour-based versus location-based), social benchmarking and real-time alerts.

For each driver, AI was used to identify priority behaviours, set achievable goals and deliver personalised coaching tailored to their driving patterns. Over four months, the system generated 23 000 personalised goals and delivered 160 000 push notifications and voice alerts (for location-based goals), as well as 390 000 coaching messages.

Johns Hopkins Bloomberg School of Public Health Associate Professor Jonathan Ehsani highlighted a clear pattern: that personalised coaching most improved driving behaviour and that improvements lasted instead of fading over time.

He also stressed the importance of collaboration between industry and academia, noting that behaviour change programmes often do not engender sustained long-term behaviour change, and in those that do, there is a wide gap between an idea and implementing it at scale in the real world.

The insurance industry is indicated to be uniquely positioned to assist, as it has access to drivers with existing communication channels, understands risk with longitudinal data about populations, and can effect behaviour change by rewarding and incentivising safer driving.

Department of Transport deputy director-general Lebohang Ntwampe says the research is aligned to South Africa’s revised National Road Safety Strategy, which was recently adapted to address the considerable numbers of fatalities on the country’s roads.

“While law enforcement remains a critical pillar of our road safety strategy, it is clear that enforcement alone will not deliver the level of change required. We must complement it with sustained behavioural change, greater accountability, and a collective commitment to protecting human life.

“Research into driver behaviour is therefore essential, as it strengthens our understanding of the factors that contribute to risky driving and helps us develop more targeted, evidence-based interventions.”

The behavioural principles tested in the study informed Discovery Insure’s ‘Advanced Driving Instructor’ - an AI-powered driving coach launched in January as a proof of concept that identifies a driver’s priority behaviour, sets personalised goals and delivers practical coaching, including location-based feedback.