Sterling Anderson doesn’t think humanoid robots should be building cars
General Motors is reorganising vehicle development around AI, centralised computing and virtual testing, aiming to cut the usual four-to-five-year development cycle to just over two years. Chief product officer Sterling Anderson argues that running thousands of virtual tests can speed design work while improving safety and coverage; he also says humanoid robots are not suited to building cars.
The plan comes as GM faces expanding Chinese electric vehicle makers, softer US EV demand and the expense of batteries and software. Anderson says GM can use its own decades of design data to train some AI systems, while drawing on outside models where appropriate. The strategy promises faster development, but depends on complex software and autonomous technologies working reliably.
In short
- GM aims to halve vehicle development time.
- Anderson says virtual tests can improve speed and safety.
- The strategy faces competition and execution risks.
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
GM’s shift to AI-led design and virtual testing could shorten development cycles and let engineers test more designs and safety scenarios before building physical prototypes. Using its own decades of vehicle data may help tailor these systems to its products, while faster development could help the company respond to competition and changing demand.
The case against
A shorter development cycle is valuable only if the software and autonomous systems work reliably; virtual tests may not capture every problem that emerges in physical vehicles or real-world use. Relying more heavily on complex software could introduce new costs and risks, so a rapid redesign of the process may not deliver the promised gains without careful validation.
Explained simply
General Motors (GM) is a major US carmaker. Developing a new vehicle usually takes four to five years, as teams design it, build prototypes and test how it performs. GM is looking to use artificial intelligence and computer simulations to shorten that process.
Sterling Anderson, GM’s chief product officer, oversees work on the company’s vehicles. AI systems can help designers assess many possible designs and run virtual tests before building physical cars. GM says it can draw on decades of its own design records to train some of these systems, alongside models developed by other companies.
The shift comes as carmakers face competition from Chinese electric vehicle brands and uncertainty about demand for electric cars in the US. Batteries and software also add to the cost of developing vehicles. Faster development could help GM respond to these pressures, though it relies on complex software and automated systems working reliably.