The Controlled Collisions of Autonomous Trucks

Waabi, a developer of self-driving trucks, takes a unique approach to testing its vehicles. Instead of conducting physical tests that could result in accidents and injuries, Waabi uses its AI-powered simulator, Waabi World, to create and replicate real-world driving scenarios. The simulator's accuracy and realism allow the company to gather valuable data without risking physical damage or harm. This data is then used to inform the development of Waabi's self-driving system, known as Waabi Driver. The company's approach has helped it progress quickly and secure investments from companies like Volvo. While other self-driving tech firms also use simulators, Waabi stands out for starting and primarily remaining in the digital realm, using simulated trucks instead of real-world vehicles. Critics, however, question the simulator's ability to reproduce genuine driving situations, and Waabi does supplement its testing with real-world experiments. Waabi is one example of a tech company using AI to generate "synthetic data," which is artificially created data that can be used in real-world applications. Another company leveraging synthetic data is Synthesis AI, which specializes in developing AI-powered facial recognition systems. Traditionally, training these systems required photographing numerous individuals, but during the COVID-19 pandemic, finding volunteers became challenging. To address this issue, Synthesis AI created a computerized 3D world where it can generate and use digital faces to train and strengthen its AI models. This approach allows for increased accuracy and enables better recognition of people with darker skin tones, addressing existing biases in facial recognition systems. The use of synthetic data also ensures privacy protection, as real-world consumer data is not needed or stored. However, some experts caution that synthetic data may not be a perfect solution and can still introduce biases if not used responsibly. Nonetheless, when used responsibly, synthetic data has the potential to create AI systems that are less biased and more privacy-invasive.