The public has become familiar with generative Artificial Intelligence (AI) as a conversational tool … [+]
Society hears two oversimplified extremes about generative AI (artificial intelligence): either it’s a fascinating language tool that responds to students’ endless demands that allow them to cheat on homework, or it’s the end-of-the-world box that leads to the Singularity when the PCs shoot. people like Sarah Connor. Most of us are not computer scientists, and so we understand AI only through pop culture lenses like ChatGPT or Terminator, thereby reducing AI to drama-inducing stereotypes.
However, this week Helm.ai, a leading provider of advanced AI software for advanced driver assistance systems (ADAS) and autonomous systems, announced the launch of VidGen-2, an AI generative model that produces twice the resolution good. than its predecessor, as well as increased realism to 30 frames per second and multi-camera capability.
“There have been various advances in AI that have dramatically accelerated the pace of development in autonomous driving,” said Helm.ai CEO and founder Vladislav Voroninski. Whatever the information content of your sensor array, an AI system should be able to make the most of it, and the only limitation is the amount of computation you apply to the problem.”
Vladislav Voroninski, CEO and founder of Helm.ai.
“Tesla has been able to invest massive amounts of capital into their fleet and the development of artificial intelligence. Other automakers have not been able to invest as much. And that means that if automakers wanted to compete with Tesla using the same approach, they would be fundamentally behind. Whereas, relying on generative AI and AI-based simulation, which helps close the sim-to-real gap, they can catch up and even surpass Tesla.”
Part of how this is achieved is through thousands of hours of training on various driving scenarios using an advanced processor from NVIDIA, innovative deep neural network (DNN) generative architectures and Deep Teaching™, an efficient unsupervised training method. . “In the earlier days, we focused mainly on unsupervised learning because it was critical to the scalable development of precision perception software and was an open problem at the time,” explains Voronisnski. “So a sort of ‘How do we train neural networks without labeled data?’ And we have made considerable progress in solving this problem [which had] has been powering our product development for several years. But recently we’ve combined that internal training technology with generative AI… to create a highly scalable version of generative AI that closes the gap between real data and simulated data.”
The result: Generative AI will reduce development time and cost while improving realism for autonomy. “Software is no longer the barrier to autonomous driving.”
Author’s note
Nearly two million of you have watched AI expert and computer science professor Stuart Russell’s TED interview, where he was essentially asked “How will AI change the world?” Russell cited both EM Forster’s stories The Machine Stops (1909) and WALL-E (2008) saying that both stories depict society becoming “weakened and infantilized” by relying so much on technology that they stop understanding how. “We have put much of our civilization in books, but books can work [things] for us, and so we have always done [taught] the next generation. If you get it, it’s about a trillion people years of teaching and learning and an unbroken chain going back tens of thousands of generations. What if that chain breaks?”
So yes, Russell is right. We all must intentionally continue to learn and then learn about learning to keep technology as our ally. These kinds of fantastic applications of generative AI are empowering, but only in a future where we continue to learn as humans as well.