Modern fleet vehicles, whether gas-powered, hybrid, or electric, rely heavily on software. From driver safety systems to navigation, entertainment, and operational controls, software is at the heart of vehicle functionality. 

Features like luxury options for comfort, advanced safety mechanisms, and essential driving systems for acceleration, steering, and braking all depend on software.

A typical new work truck contains more than 100 million lines of code, underscoring the importance of software quality in every model produced by General Motors (GM).

Prioritizing Quality Over Experimentation

Unlike other industries, where software testing may rely on user feedback during beta phases, General Motors holds itself to a higher standard. The company’s quality labs are dedicated to identifying and resolving potential software issues well before the code is deployed in customer vehicles.

GM’s software team includes engineers with experience at leading tech companies, drawn by the opportunity to tackle the complex challenges involved in modern vehicle development.

Leveraging AI for Around-the-Clock Testing

AI plays an increasingly critical role in GM’s approach to software development. Automated systems perform continuous testing of vehicle software, simulating years of real-world use to ensure reliability and quality.

For example, software for dashboard touchscreens undergoes rigorous virtual testing at GM’s quality labs. These tests simulate millions of button presses and screen taps, helping identify bugs before human testers or drivers interact with the software.

Simulated and Real-World Testing Integration

After virtual testing, engineers conduct real-world tests at GM’s Milford Proving Ground in Michigan and other locations. Some vehicles in these trials are equipped with custom devices that automate software testing even while parked. Machine learning-powered tools continue the testing overnight, refining the software for more consistent and predictable performance at release.

Automated and simulated testing significantly reduces the time required compared to physical tests alone. Virtual testing is bolstered by real-world data gathered from GM employees who conduct test drives across the U.S. to ensure the software performs in varied driving environments.

A Global Network of Testing Facilities

GM’s commitment to software quality is evident in its extensive testing infrastructure. The company operates labs in Michigan, California, and Ontario, Canada. One Michigan facility houses over 300 test benches dedicated to debugging dashboard infotainment systems.

AI and machine learning enable GM to run increasingly sophisticated tests, continuously improving the software that powers its vehicles.

The automaker is also embracing renewable energy, singing a 15-year renewable energy purchase agreement with NorthStar Clean Energy for three GM assembly plants as part of its goal to power all production sites with renewable electricity by the end of 2025.