Last month, during the company’s earnings call, GM CEO Mary Barra updated the world on the company’s autonomous driving ambitions.
General Motors is working on eyes-off, hands-off self-driving technology that will launch in 2028 on the Cadillac Escalade IQ. Such a breakthrough would put it ahead of what Tesla currently offers with Full Self-Driving (Supervised), as FSD is an eyes-on, hands-off system.
But GM’s technical wizardry didn’t end there. Barra also said the company was testing the new autonomous driving technology “in a digital environment” that allows the company to simulate about 100 years of human driving in a single day.
While GM’s testing isn’t limited to the digital world (the company also began on-road testing in California and Michigan), Barra did emphasize that GM is fully embracing AI “across the enterprise.”
“Today, nearly 90% of the code written by our autonomy team is generated by AI,” Barra boasted. The moves are part of a larger shift across the entire company to rely more on artificial intelligence to reconfigure the entire way it looks at vehicle development.
GM wants to transform the rules of vehicle development
Developing a new vehicle is an arduous process that involves many interdependent disciplines working separately to achieve a singular goal.
GM likens that old system to a relay race where each team advanced the work they were assigned, then handed that work off to the next team so they could do the same. While that approach has worked for over a century, the company says there’s an inherent flaw.
If one team drops the baton, it could spell doom for the entire project.
So the automaker is relying on artificial intelligence and virtual labs to allow teams to work individually, but simultaneously. “GM is building something different now: a concurrent development system wherein more teams can work from the same digital foundation at the same time,” the company said recently.
“Designers, aerodynamicists, software teams, controls engineers, safety specialists, and manufacturing experts can increasingly learn from the same model, allowing them to react faster to changes and make better decisions earlier in the process.”
GM says it is switching its process from handoffs to “concurrent engineering.” So instead of one team working on one thing until they are finished, GM can have multiple teams working on different things at the same time for the same vehicle.
“This is bigger than any one AI tool,” the company says. “It is a shift in operating model, evolving from disconnected phases to a single integrated digital thread spanning design, engineering, validation, and manufacturing. In a software-defined era, that kind of zero-lag iteration is becoming a competitive advantage.”
Tramino / Getty Images
What does GM hope to achieve with its AI-led process?
General Motors says its new design approach will allow different teams to work in parallel, allowing for faster feedback and fewer instances of “late-stage surprises.”
GM confirms it will lean heavily on AI, embedding it into workflows from the earliest stages of development. Still, human creativity “sets the direction while AI helps expand design-space exploration by making it possible to generate more variations, evaluate them earlier, and focus more time on the choices that matter most.”
The carmaker is also using co-simulation or “co-sim,” which allows designers to validate how the software they are building interacts with the hardware it’s meant for, even before that hardware is ready.
GM said its new AI and virtualization approach is “becoming a core competency, and a new way to imagine, design, develop, validate, and manufacture better vehicles for the future.”
Related: General Motors analyst spots customer trend investors should watch