The automaker's aggressive AI adoption is reshaping how its engineering teams write, test, and ship code at scale
General Motors is going all-in on AI-assisted software development. The automaker says it achieved a 300% increase in merged pull requests after retooling its AI agent software, a metric that signals dramatically more code is being reviewed, approved, and integrated into production systems.
For the non-engineers in the room: a “merged pull request” is what happens when a developer’s proposed code change gets approved and folded into the main codebase. A 300% jump means GM’s teams are shipping four times the volume of completed code contributions they were before.
GM’s AI coding machine #
During the company’s Q1 2026 earnings call, CEO Mary Barra disclosed that nearly 90% of the code produced by GM’s autonomy team is now generated using AI tools.
GM has also invested heavily in automated testing infrastructure. The company employs AI-driven testing labs that simulate millions of user interactions on vehicle interfaces, helping validate software quality and safety before updates ever reach a customer’s car.
These testing capabilities feed directly into GM’s Ultifi platform, which enables continuous over-the-air software updates across the fleet.
What this means for investors #
GM’s Super Cruise advanced driver-assistance system already positions it as a technology leader among legacy automakers.
The risk side deserves attention too. Heavy reliance on AI-generated code at this scale is relatively uncharted territory, particularly in safety-critical automotive applications. GM’s 90% AI code generation figure for its autonomy team is impressive, but it also means the company is betting heavily that its validation processes can catch what its AI misses.
Investors watching this space should track not just output metrics like pull request volume, but downstream indicators: over-the-air update frequency, Super Cruise expansion milestones, and any safety incidents that might signal quality gaps in the accelerated pipeline.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our