“There will be a before and after of generative AI”. Designer of the Siri voice assistant and
now scientific director at Renault, Luc Julia is at the forefront of the battle for artificial intelligence in the automobile. If AI has been cited on numerous occasions to revolutionize transport, notably thanks to autonomous carsshe is gradually interfering in the industry.
According to a study by the German firm Roland Berger, in Research and Development (R&D) alone, it would allow equipment manufacturers, specialists in a specific automotive part, to reduce their spending by 30% to 50%. And this is thanks to the time saved to carry out a task, up to 3 years out of the 5 years needed to develop a new product. “ In the innovation phase, we find in particular the crucial step of reading patents which makes it possible to map the state of the technological art and its intellectual property. The execution time today can approach 10 days. Thanks to AI, it will be able to drop to 1h30”confirms Clément Carle, senior consultant of the firm.
Saving time and money can be done on several levels. Concretely, generative AI can generate new part designs based on the instructions given by the car manufacturer, while improving its aerodynamics, its resistance, etc. Above all, it makes it possible to accelerate the testing and validation phases and, thus, improving product quality. “ The AI will consult all the old test reports and adjust their number by justifying its choice in relation to the previous performances that it has recorded in its database.explains Clément Carle.
In addition to increasing possibilities, AI will mainly affect jobs. “In principle, AI develops more when it can replace several jobs, preferably in the service sector”, says Cyrille Dalmont, director of research on digital issues at the Thomas More Institute. This is also the reason why AI is only in its infancy in the automotive industry, unlike banks, insurance or even the commercial sector.
Each company must therefore create its own ChatGPT, trained according to the group’s historical data and its objectives. For the moment, it is difficult to know the progress of these systems as well as their performance in the field, as the industrial secret is so well guarded. But the car manufacturer that manages to generalize AI within its production will have a considerable head start over the others.
France very late
And everything suggests that Chinese, Korean and American car manufacturers are in the best position. “France is very late since in Europe, we do not have world-class digital giants”, recalls Cyrille Dalmont. In this battle, Tesla, created entirely around digital technology, appears to be the most advanced competitor. For its part, China has announced its ambition to be the pioneer in digitalization, in particular in autonomous cars. But nothing on the industry side.
Because artificial intelligence in vehicle production requires colossal resources, and car manufacturers are busy in other areas. In particular the Europeans, who must manage the imposed shift towards electric, with the end of the sale of thermal engines by 2035.
Likewise, European manufacturers prefer to focus on digitalization on board the vehicle. Recently, Stellantis announced the implementation of ChatGPT in its future DS4 marketed in 2024. A European first which shows the desire of historic manufacturers to enter the AI battle. Especially since, recently, generative AI has become widely democratized with open source language models, allowing manufacturers to have their own AI without being linked to GAFAM. These systems make it possible to keep manufacturer data within the group, and thus reduce the risk of cyberattacks. A major breakthrough, particularly in the very confidential and strategic world of industrial production.