The field of AI research is thought to have originated at a workshop held on the campus of Dartmouth College in the summer of 1956; it took almost 60 more years for the technology to find a place in everyday applications.
AI, in the form of machine learning, has been used by search engine companies, such as Google and Yandex, and autonomous vehicle leaders such as Waymo and Tesla, but it was not until February 2023 when ChatGPT was released to the public that generative AI went mainstream.
With large-language models (LLMs), such as ChatGPT, proving that machines can replicate normal human conversation, the automotive industry examined how this tech could be offered to drivers and thereby unlocked its share of the $700 billion AI market opportunity. By allowing intelligent digital voice assistants to be integrated into vehicles, manufacturers offer drivers a simpler human machine interface (HMI) with the vehicle systems. These voice assistants use natural language processing (NLP) and speech recognition technology to understand and respond to the driver’s or passenger’s voice commands.
A personal assistant in your dashboard
Enabled by AI and vehicle-to-everything (V2X) connectivity, and underpinned by the software-defined vehicle (SDV) architecture, in-cabin connected services and assisted driving systems are rapidly approaching maturity.
This was practically demonstrated at CES 2024 where Qualcomm and Amazon Web Services (AWS) revealed Snapdragon interfaces that had been trained on the vehicle's owner manual. In these demonstrations, customers were able to ask about the features of the vehicle and receive a logical explanation from the AI assistant.
Voice and gesture controls in automotive environments can be problematic, in a noisy passenger cabin where the driver needs to maintain utmost attention to the task at hand. While many cars are already equipped with voice control systems that interface with the vehicle’s systems, AI assistants offer a far more natural way of doing this. Drivers will not need to hone their commands and use specific words; instead, instructions can be issued using natural language.
What is more, in an environment of shrinking aftersales revenue, OEMs and their dealers will be able to leverage the enhanced UX provided by these platforms to boost subscription services. AI assistants will mostly be premium paid-services, perhaps with the first year free to demonstrate value and attract customers. For instance, AWS's AI assistant for the car is a pay-for-use cloud service that connects to AWS servers to execute its AI functionality. And, to ensure a seamless interface for critical applications at all times — even when the connection to the cloud fails —Qualcomm's latest chips incorporate built-in AI accelerators, allowing the vehicle to support certain AI features even when offline.
The AI cabin assistant also need not be limited to in-cabin interactions with the vehicle — combined with connectivity and third-party platforms, the functional scope of the AI assistant is almost endless. For instance, similar to GM’s OnStar connected service, the AI assistant could perform familiar tasks such as scheduling car servicing. However, in this case, when the vehicle gets close to its service interval, the application would access the dealership's service center availability, then offer the driver available slots to get their vehicle service, let them know the cost of different packages and even schedule and pay for the appointment
This seamless in-vehicle payment functionality, powered by AI, promises to be a game-changing technology for automakers and service providers. Merely stating "I'm cold" would be enough to spur the AI assistant into action — no button presses or digital interfaces to cope with. In response, the assistant could increase the cabin ambient temperature, or upsell the driver on heated seats or other unlockable upgrades.
Connectivity and AI payments, then seamlessly facilitate the transaction. Tesla and BMW offer similar features today, although the AI isn't responsible to process payments. To ensure the transaction is secure, AI systems will need to consider security measures such as biometric recognition powered by the in-cabin infrared and regular cameras, to authorize safe payments. And blockchain ledgers to keep financial information secure.
VW could be first to market
Recognizing the potential of generative AI in enhancing the user experience, Volkswagen is set to become the first volume manufacturer to offer voice-activated, AI-based, driver assistance software as a standard feature in vehicles from the compact segment upwards. The company chose to base the service on the ChatGPT platform.
The inclusion of an AI assistant, accessed via a “Hello IDA” voice control, unlocks a range of novel capabilities for owners of several upcoming VW models. From navigation to managing infotainment systems, getting weather or traffic updates, and answering questions regarding destination and immediate surroundings, ChatGPT’s NLP unlocks all these possibilities through integration with Cerence Chat Pro.
Partnering with Cerence Inc., gives VW seamless access to automotive-grade ChatGPT integration, offering unmatched system flexibility, customization and ease of deployment, while ensuring the consumer’s security. Crucially, ChatGPT in this application does not access any critical vehicle data, while all user questions and system answers are deleted immediately to ensure the highest possible level of data protection.
Looking to the future, Volkswagen promises that, together with Cerence, it will continue to explore new LLM-based user experience opportunities as the foundation of Volkswagen’s next-generation in-car assistant.
Conclusion
AI-powered conversational assistants will enable drivers to access car and connectivity features in a way not previously possible. This will enable users to literally talk to their cars — think KITT from Knight Rider, but perhaps less mordant.
But are consumers clamoring for AI assistances in their cars? A recent study of consumer drivers found that about half are interested in adding generative, communicative AI to their in-car experience. So there is clearly market appetite.
But less clear are the downsides. Perhaps the most important might be, how it could serve wrong information — known as AI hallucinations — which in an automotive environment could prove deadline. An AI presence might also change human relationships with cars, which is a complex culture. Digital platforms are often prone to platform decay, wherein the increased pressure to monetize every experience leads to decreased user happiness. Could AI assistants change the primitive mechanical connection humans have with cars? Maybe.
As with many applications, the impact of generative AI on this industry is speculative. But with the ongoing R&D of VW and other automakers, drivers will no longer be talking to themselves when stuck in traffic.
About the author
Peter Els is a South Africa-based former automotive engineer. This includes time with Nissan South Africa’s Product Development Division, Daimler Chrysler, Toyota, Fiat/Alfa Romeo, Beijing Automotive Works (BAW), as well as tier-one suppliers Robert Bosch and Pi Shurlok. After consulting to the local industry for 15 years, Els has ventured into technical writing and journalism about the latest trends, technologies, opportunities and threats facing the new world of mobility.
