From Alexa to Nuance, including personal assistants who are learning your mood, an overview of the AI-based features that major automotive companies plan to deploy soon.


AI assistants by car arrive whether you like it or not

Aurich / Getty

LAS VEGAS – Whether we like it or not, CES has become an auto show, for the same reason we cover the world of automobiles at Ars Technica. In simple terms, the technology sector has looked at the automobile and found dollar signs. Whether or not this large annual trade show is the best way to start a new year (no problem – that's not the case), attending CES is an asset to spot trends. And this year, the main trend seemed to be "the same product as last year, but with the AI": AI compatible TVs, induction cooktops compatible with the I & # 39; AI and, yes, AI in cars.

In fact, the idea of ​​a personal car assistant in artificial intelligence has been around for quite some time. I had a first glimpse of this new world in 2016 when Audi introduced its concept called PIA (for Personal Intelligent Assistant). Since then, I have heard about these aids to AI from a growing number of auto manufacturers, and the technology is getting closer to production.

Take BMW. In just a few months, in some markets you will be able to purchase a 3 Series (or 8 Series, or X5 or Z4) model including the company's new smart personal assistant, which is part of the new Seventh Generation Infotainment System. By operating a pretty good voice recognition (greet Nuance), you can give the car instructions such as: "Hey BMW, I'm cold," at which point the cabin temperature will increase for you. "We expect a much greater commitment to voice interaction," said Dieter May, BMW's senior vice president of digital products and services.

May is far from being the only one in her intuition. Audi, Chrysler, Daimler, Ford and Hyundai also use the Nuance Dragon Drive platform to allow us to talk to our cars (and make us understand). And this is not the only game in town; Amazon and Google are also making inroads in the automobile, often with the same builders. For example, you can already buy BMWs and Toyota with Alexa integration and Volvo adds Google Assistant to its infotainment system. For the smart home user, this means that you can ask, for example, if your garage door is still open or if you forgot to turn off the lights when you left the house.

As we expand the lead time, the AI ​​in the car promises to do a lot more. Byton, a new electric vehicle start-up that is expected to launch its first vehicle in 2020, promises that its electric vehicles will use facial recognition as biometric authentication, snatching your Byton profile from the cloud. But "in the process, it fully adapts to your needs, welcoming you personally as it automatically adjusts the seats, the temperature and the preferences of the screen".

And you can bet that all the driver monitoring systems we've seen will transmit this information to the onboard AI that uses it. In addition to the warnings given when he thinks he is sleepy or distracted, you may be asked if you want to hear soothing music if the system thinks you are preparing for road rage. (It probably means that your future car will know when you swear it too.)

Who asks for this?

What is the fault with the technization (that's a word, honest) cars? This is perhaps the Tesla effect. When the Model S arrived with a large touch screen and updates over the air, people began to realize this, and now every OEM will tell you that customers are calling for an experience more similar to the one they have. a smartphone from their vehicles. This is why high-resolution displays are replacing analog gauges, the old-school (and unsecured) CAN bus is joined by Ethernet Ethernet, and powerful GPUs are moving from gaming platforms to domain controllers that will consolidate them. dozens of discrete black boxes. currently running each of the many numerically adjustable attributes of a car.

In each case, cars represent a lucrative new market for the consumer technology sector. For example, a company like Aquantia sells well Ethernet solutions to data centers and sites such as Apple. But these traditional business sectors could be negated if it is the one that provides multi-gigabit Ethernet systems to millions of new cars every year. And the same goes for companies such as Nvidia, Qualcomm or Intel. All this sophisticated new equipment needs a reason to be there, after all.

Some applications seem really useful. A car that knows when it is about to repair sound seems very useful, for example. As well as being able to easily search directions. And I used BMW 's current voice recognition system to interrogate the user' s manual more than once when testing some of his cars. According to Dirk Wollschläger, general manager of IBM for the global automotive industry, AI will ensure that your car, for example, does not ask you to apply a critical update when you are in the middle of your travels daily.

"For this type of use case, we also use artificial intelligence – we only want to convey the relevant information to the driver – the end user needs to be able to define when they want this information." should not receive an update request if you are busy driving, but if you park, he knows that you are now parked and can then give you a prompt or something else.We are working with OEMs there " said Wollschläger to Ars.

But all too often, these artificial intelligence systems are presented as an additional way to get us to buy things, but this time on the move. And knowing the public is the last thing we all want. But maybe I'm too cynical. Upton Bowden, Director of Advanced Technology Development at Visteon, thinks I may be. "In our study, we found regional differences regarding artificial intelligence assistants.The US is fascinated by the memories of Microsoft's paperclip, I think.While in Asia, they see things differently and are willing to have this feature, better it will lead to use, "he told me.

Update: The Alexa team has communicated with statistics showing that many people claim these features, even if they are poorly represented in the comments here in Ars: "The Alexa Auto team has just completed a study with JD Power interviewed by consumers already owning a speaker using the voice, like Amazon Echo, and having found that 76% of people who interact with a voice assistant at home said they want the same assistant in their car, have already bought more than 100 million people. Alexa-compatible devices and you're starting to get an idea of ​​the demand. "