Waabi is building Level 4 autonomous trucks using verifiable end-to-end AI systems, a stark departure from the black box architectures dominating passenger car automation.1 CEO Raquel Urtasun stated that Level 2+ passenger car systems rely on unverifiable neural networks that cannot support true autonomy.2
The company's Waabi Driver system combines computer vision with decision-making in a transparent architecture that allows engineers to validate behavior before deployment. Snowstorms still create operational no-go zones for the platform.3
Urtasun's approach addresses growing concerns about deploying AI systems that cannot be audited or explained. The autonomous vehicle industry faces 2 million global road deaths annually, creating pressure for provably safe systems.4
The debate over verifiable versus black box AI extends beyond autonomous vehicles. Meta's Yann LeCun recently argued that no individual—including himself, Dario Amodei, Sam Altman, or Elon Musk—has legitimacy to unilaterally decide acceptable AI applications for society.1 This philosophical divide reflects practical engineering choices: transparent systems that can be validated versus opaque neural networks optimized purely for performance.
Waabi's stance suggests the autonomous trucking sector will prioritize interpretability over the deep learning approaches that powered consumer vehicle features like lane-keeping and adaptive cruise control. Urtasun believes current truck drivers will be able to retire in their profession, indicating a gradual deployment timeline.2
The technical challenge centers on creating end-to-end vision-language-action models that maintain verifiability while handling highway driving's complexity. Unlike geofenced robotaxi operations, long-haul trucking requires systems that navigate diverse weather, traffic patterns, and infrastructure across thousands of miles.
Whether verifiable architectures can match black box performance remains the industry's central question. Waabi's approach bets that transparency and safety certification will outweigh any performance gap as regulators and insurers evaluate autonomous commercial vehicles.