Years from now, cars will drive themselves. But car owners won’t have to wait that long to enjoy the benefits of learning algorithms. We already have the tools to create digital assistants to handle everyday car-owner responsibilities.
The key to making this happen is orchestrating a series of AI engines that use predictive capability, pattern matching, natural language processing (NLP) and other machine-learning methodologies. Consider a straightforward use case: Your car needs an oil change.
Depending on the model, your car may be able to read your odometer and turn on a light telling you it’s time to change the oil. But you still have to contact the dealer and sync your schedule with available times in the dealer’s garage. If you use premium synthetic oil, you have to tell somebody in the service department and follow a specific schedule.
The next phase of AI in automotive is to use learning algorithms to automate much of this process. Several AI engines could be put into play:
- A predictive algorithm can draw data from multiple sources to recommend the optimum mileage for changing your oil. The schedule accounts for your favorite synthetic oil.
- A digital assistant algorithm can scan your calendar and sync it with available garage slots at your car dealer.
- A pattern-matching algorithm can check all the previous times you’ve taken your car in for an oil change. If you usually take the car in on Tuesday mornings, the AI starts by looking for open dates at the dealer on those days and times.
- An NLP algorithm can power a voice interface that walks you through a series of questions and options to set up the appointment hands-free while you’re driving to work.
The real value here doesn’t come from creating the algorithms. It comes from coordinating them to produce the desired outcome. This principle holds for any AI-powered services for car owners. As in-car services evolve, OEMs and vendors will have to figure out ways to simplify the lives of car owners — or stand by and watch somebody else do it.
Challenges to Improving the Car-Owner Experience with AI
Whether you’re an OEM or vendor in the automotive sector, you feel the pressure to use AI to improve the owner experience. Everybody has the same question: Where to begin? The best route is to start small: Develop one AI project that works. Find complementary areas where AI makes sense and delivers value. Build on what you’ve learned in the earliest phases.
The greatest mistake is taking on too much, too soon.
Another sticky challenge is data silos. Dealers, manufacturers and suppliers all have datasets that might not be compatible. Everybody has data-governance and security concerns. Compliance issues may cloud the picture even more. Getting all these factions to share their data is a demanding job.
But to make AI work at scale in a large industry like automotive, you have to knock down these silos.
Driving outcomes with data
If you have enough data, you can create an algorithm that strives to achieve a desired outcome. The algorithm is optimized to produce wins and avoid losses. Over time, the algorithm accumulates so many wins that it learns to perform better without human intervention.
In the auto sector, effective AI means using data to capture the ownership experience — and make it better. Data from smartphones, apps, and transactions helps us understand a car owner’s behavior and develop services that remove friction from their lives.
Putting data to work in predictive algorithms is the critical first step. But to drive lasting impact for car owners, you have to coordinate multiple AI engines to create a cohesive experience.
That’s the outcome we’re driving at DMI.
-Niraj Patel, senior vice president, artificial intelligence