The automotive industry is increasingly focused on artificial intelligence (AI), but progress isn’t just about chatbots or image generators. Major players like Bosch and Volkswagen’s Cariad are investing heavily in AI to create self-driving systems that mimic human driving behavior.
The Automated Driving Alliance: Levels 2 and 3 Autonomy
In 2022, Bosch and Cariad formed the Automated Driving Alliance with the goal of developing autonomous systems for levels 2 and 3. Level 2 allows hands-free driving on various road types, while level 3 enables total vehicle control on highways. This is more than just automated features; it’s about building systems that can genuinely think like a human driver.
Beyond Data Analysis: Cognitive AI in Vehicles
AI isn’t monolithic. While it’s well-known for content generation, its deeper application lies in analyzing complex data more effectively than traditional methods. Engineers are already using AI to accelerate the analysis of test data during car development, speeding up innovation. Bosch and Cariad aim to go further, creating AI-based software that can perform all essential cognitive tasks: perception, interpretation, decision-making, and action.
The Pursuit of “Cognitive” Driving
To replace a human driver entirely, a system must not only analyze situations but also reason and make critical decisions. The goal is a vehicle that can learn, understand, and act independently. This means the AI must handle basic tasks like knowing its location and traffic rules, but also more complex scenarios that require judgment.
Vision-Language-Action: The Next Step
The current software lays a foundation for integrating “vision-language-action” approaches, which would allow machines to mimic human logical thinking. This would enable cars to assess hidden risks in traffic situations, just as a human driver might. The key is that the AI’s actions must be safe, traceable, and explainable. Accountability is essential.
Production Timeline and Continuous Improvement
Bosch and Cariad’s AI-based software stack is currently being tested and is expected to be production-ready by mid-2026. The system will continuously improve through massive data input, making automated driving at levels 2 and 3 more robust over time. This isn’t just about automating tasks; it’s about creating a system that learns and adapts like a human driver, but without the flaws of human error.
The ultimate ambition is to develop autonomous systems capable of acting “as naturally as the human driver,” which represents a significant leap forward in automotive AI.
