Artificial intelligence (AI) and intelligent technologies in general are inseparable in our daily lives and because of this broad spectrum, AI is often referred to as general purpose technology. It may take time until steering wheels and pedals are no longer necessary, but AI already plays a significant role in so many other contexts within the automotive sector. From production planning to design and development, and from logistics to sales, nearly every aspect of car manufacturing and distribution is or will be enhanced in the mid- and long-term through the use of AI.
The automotive industry is facing the dawn of the second machine age. Let’s take a look at how AI has changed and will change the automotive industry.
What comes after Big Data? Better Data!
AI is a double-edged sword—on the one hand, there are endless possibilities to automate complex processes; on the other, the crux lies precisely in this complexity, since in order to carry out highly sophisticated tasks, a high-quality database is required. The quality of the database accessed by AI, regardless of the context, is not determined by data density. A data-heavy project usually involves post-processing logic, ETL processes, various analytical and cognitive components and data streams.
It may take time until steering wheels and pedals are no longer necessary, but AI already plays a significant role in so many other contexts within the automotive sector
The most critical part of such a scenario is a high-performance data processing pipeline, which feeds and maintains a data store, which briefly describes of a data environment that enriches analytic models, AI applications and decision-making in real-time. Such actions are critical when it comes to AI, since they are the base of any kind of AI-driven solution.
What’s the deal for the automotive industry?
Think of the automotive sector and consider the ‘scope of relevance’ that a car manufacturer has these days: it is limited to the car, but most of an automaker’s customers spend only a small amount of their time in their car. Automakers therefore need to expand their footprint way beyond the car, and into the world of mobile devices in order to become a natural part of their customers’ lives. To achieve this presence by using AI, trust is a vital asset, and as AI grows in its capabilities—and impacts people’s lives—businesses must move to ‘raise’ their AI to act as a responsible, productive member of society. Nearly 80% agree that within two years, Al will become an ever-present ‘co-worker’ and trusted advisor to humans in various aspects of their lives, including for example their next car purchase.
At this stage, we are far from talking about proprietary AI and machine learning processes, and there is little evidence to deliver highly reliable results when it comes to complex processes. However, let’s take a look at what is already working well, and also consider the use of AI in the medium and long term.
NLP—natural language processing: There are many gaps between a car and its driver, and one way to bridge the gap is to develop natural language processing: the ability of an infotainment system to recognise and process natural language. Think of questions such as, “How do I open the hood?”, “What is the current oil level?”, or “Which warning lamp is flashing?”. Embedding the car manual via an XML file, for instance, and linking it to an NLP system could be a resource-saving and functional base for proper communication between a car and its driver. And automakers can take advantage from this derived data: someone asking about the oil level might well need a vehicle inspection or an oil change.
The more precisely a vehicle can identify its driver, the better it can adapt to their preferences. On the user side, this means less effort, less distraction and improved safety
Identify in-car needs by AI: A second useful AI application is to identify driver or passenger profiles, something which is way beyond simple key recognition. The example of automatic climate control illustrates how a vehicle can learn to use data analysis. Simple data, such as how many passengers are in a car and where they are sitting shows the car where it needs to spread heated or cooled air, or activate heated seats.
Based on a manageable amount of data, AI can be used to create typical driver profiles: Does the driver’s behaviour with regard to seat heating change depending on the time of day? What influence does the outside temperature have on the heating level?
The more precisely a vehicle can identify its driver, the better it can adapt to their preferences. On the user side, this means less effort, less distraction and improved safety. Using machine learning, data can be used to identify different input patterns and assign drivers. For automakers, these offer financial and functional advantages in terms of bandwidth and computing power for low-resource AI models.
Cross-selling and up-selling—driven by AI: “If you liked that, you might also like this”—such is the way e-commerce businesses seek to recommend supplementary products based on our previous purchases and online browsing behaviour. Certain e-commerce players have enjoyed double-digit increases in terms of sales by embracing cross- and up-selling features. Think of car configuration and imagine that an automaker recommended you to add underbody protection, the online configurator having recognised from your previous web searches that you are interested in off-road driving; this is just one of countless examples of how automakers can enhance the online experience of existing and prospective customers, while at the same time increasing their revenues.
AI is already able to make a significant impact on the automotive industry value chain by affecting every part of the business—and its impact is only set to increase
Higher quality and efficiency, but with less effort: Sounds like a dream, right? It isn’t: combining AI and predictive maintenance can be powerful when incorporated into the production process. A predictive maintenance system enriched with comprehensive AI data is able to identify anomalies within the painting process, for example, thereby enhancing quality. Or think of certain car parts, such as a door: using AI, a manufacturer can predict how often a door is opened and closed, and from that figure derive the necessary quality for a door hinge.
It’s all about the right quantum—predictive mobility: Picture the scene: you are just heading from the Champs-Élysées to the taxi stand, along with five strangers; six taxis are needed, and surprisingly, there are six taxis. Coincidence? Not at all—at least, not in the future. By embracing quantum computing and other cutting-edge technologies, mobility providers such as taxi companies, car-sharing and ride-hailing companies, and providers of navigation services, are able to predict traffic volume and the need for mobility solutions.
Quantum computing is able to forecast the demand for transportation, meaning that everyone will get from A to B whenever and wherever they want. Furthermore, a feature that can predict traffic jams will ensure that customers will not only get the kind of mobility they are looking for, but they will also experience it in comfort.
AI will impact every aspect of the automotive value chain
The bottom line is that AI is already able to make a significant impact on the automotive industry value chain by affecting every part of the business—and its impact is only set to increase. AI will enhance quality, increase efficiency, decrease risk and, driven by data, AI will improve the type and level of service that automakers can deliver to their existing and prospective customers, ensuring that everyone benefits from the best possible product.
Axel Schmidt is Senior Managing Director, Industry Managing Director Mobility at Accenture
This article is taken from Automotive World’s December 2019 ‘Special report: how will artificial intelligence help run the automotive industry?’, which is available now to download
