Our Subaru Outback, 10, manages winter conditions like Lindsey vonn, and its adaptive cruise control system at sight is a boon on long journeys, but the same trip to UTAH a few years ago let me feel drained.
Since then, I wondered: could the new technology make the trip feel less like a Griswolds road trip and more as a trip to the East Express? And it almost did it.
But it’s always a 12 -hour period when you can’t do a lot. The next challenge for the automotive industry is to make driving systems so good that it is sure on certain sections so that drivers are leaving your eyes.
Why we are not here is already a bit of a scraper-below while GM technology was good enough to offer customers taxi trips around San Francisco without a person in the driver’s seat.
Another type of challenge. Components must be mass produced at reasonable prices, with incredible reliability levels. And car manufacturers cannot sell ADAS systems that only work in a city. The targeting of highways, which requires fewer complex decisions, is logical. But the consequences of a computer problem when traveling at 75 MPH are high. Tesla drivers who trust technology too much (ignore or get around safety measures) paid with Their life.
The AI revolution in recent years is complicated the question. Systems like Super Cruise are guided by an inflexible Meli-Mélo software puree by design. As far as possible, the company tried to make the car as if it were on digital rails. There is no chance that he hallucinates and suddenly decides to go to traffic in the opposite direction – although he made mistakes along our route which required corrective measures.
But some autonomous driving companies implement AI models that operate using the same type of architecture that feeds Chatgpt – known as transformers. Google, for example, has experienced the use of its Gemini AI models to operate its Waymo driverless taxis.
The Waabi autonomous trucking company uses generative AI models in a photorealist simulation of the real world for its training.
Adam Rodriguez, who runs a product for GM ADAS systems, said the company had also experienced transformer architectures.
The Tesla’s autonomous software 14 at all is According to rumors, employment A transformer architecture. If it works, it could upset the autonomous driving race and give Tesla an even larger step ahead. Tesla CEO, Elon Musk, has announced that the company will start to offer a driver -free taxi service in Texas and recently started The regulation process for one in California. (Musc deadlines are at best guest of what will happen).
If the transformers are the future, Tesla would have another advantage. The models of foundations based on transformers require massive quantities of power and calculation data, and Musk also controls XAI, which has the largest group of ia chips in the world and has just published a border model that competes with the best in the world.
But unless IA breakthroughs suddenly accelerate, which is always possible, it is likely that it will take years before the foundation models are reliable enough to operate cars without human supervision on any roadway (called level 5 autonomy).
There may be an intermediate period when the most advanced AI is not necessarily the best consumption experience. This opens an opportunity for GM and other car manufacturers to be the first to sell cars that reach what is often called “level 4” autonomy, allowing drivers to take a nap or work on a laptop while the car moves.
And if that happens, I may not be the only person who wants to drive 766 miles instead of taking a two-hour flight.