Artificial intelligence, here to help?

Lately, everyone seems to be talking about the latest source of excitement and even concern. AI, ‘artificial intelligence’, is an umbrella term that covers a vast array of things that all have a point in common: they involve machine assistance for human-specific tasks. Renault Group has experience in human-machine interactions, so what can AI bring to a ‘next-generation automotive company’? How is it going to be developed within the Group to better serve employees?

From fiction to reality

While the term ‘artificial intelligence’ tends to evoke images of HAL 9000 (2001: A Space Odyssey) or OS1 (Her), it actually refers to something a little different. AI is not about sentient machines with free-will who may just bring about our very annihilation. Rather, AI is a set of computational and mathematical models that harness considerable computing power to perform complex tasks that are otherwise impossible or to greatly optimise existing processes.

“AI has been used at Renault Group for many years,” confirmed Rodolphe Gelin , Expert Leader IA, Renault Group. “The most visible example was the ADAS on-board camera, which completely changed the way people drive. But we also use an array of AI in many of our industrial and engineering processes. So, it’s not a revolution but an evolution.”

There is a distinct correlation between development seen in the AI sector and the rapid growth of data. The past decade saw a steep rise in the volumes of collectable and collected data. Data is everywhere and everything can become data. “In 2017, when we took a gamble on Industry 4.0 by connecting sensors to machines in our factories to extract real-time data, we suddenly began producing colossal amounts of data that only AI could process and use,” explained Anthony Vouillon , Data Science Chapter Leader at Renault Group. “The Group’s approach is to identify those value drivers, processes, uses, and projects that can benefit greatly from AI for the very reason that they are powered by vast amounts of data.”

AI: already here and ready

AI is already being used by the Quality division. “For Quality work, AI’s main contribution is in accident analysis,” continued Anthony Vouillon. “For example, language processing helps better qualify customer testimonials to identify weak signals and optimise action plans. As such, we can better detect a fault or issue during the design phase.” On the industrial side, the Industry 4.0 plan boosted predictive maintenance capabilities thanks to real-time data feedback loops.

Above all, AI has made it possible to improve the day-to-day work of production line operators: “For example, according to WLTP cycle [editor’s note: Worldwide Harmonised Light Vehicle Test Procedure] requirements, we have to check that the left and right tyres are exactly the same in terms of certain criteria,” explained Anthony Vouillon. “This time-consuming task used to be performed by an operator; today, AI is used to immediately compare photographic reference points, and to raise an alarm should the tyres not be compliant. The operator’s time can therefore be focused back on tasks with higher added value: 80% of their time is now spent correcting rather than identifying.”

This vision of the ‘AI-augmented operator’ also helps foster a more positive image of industrial professions. Without an operator, a system's feedback loop does not work, it requires intelligent input from the operator for the model to grow and improve. It is not a question of ‘man OR machine’, but ‘man AND machine’

AI can also help with decision-making, like that in after-sales services: “We have fairly good visibility on what parts are needed by local sales teams for those partners who are well-integrated into our IT systems (which we refer to as R1 and R2),” explained Anthony Vouillon. "That said, for R3 — those partners who are far removed from our systems — it is impossible to know precisely what they need as they are entirely independent from our own systems. AI allows us to create predictive models, based on the behaviour of R1 and R2 partners, to determine what parts R3 partners were likely to need, thereby giving clearer insights to our sales representatives.”

There are other, less-exploited fields, with great AI potential: “In terms of human resources (HR), for example, we are working to streamline internal mobility. AI models could help us set up a system to match everyone's skills with open positions, thereby being more proactive when it comes to career management. We have also started using AI to predict office occupancy rates with the rise in remote work. This means we can better manage resources by optimising office space,” said Anthony Vouillon.

All-intelligent, but not at any old how

Throughout the Group, 150 brand-new projects and initiatives have already been identified as likely to benefit from AI. Instead of just rolling out the technology, a great deal of thought is going into doing it in a structured way. "Our aim is to make sure that everyone who has to do AI does it, as long as they do it well," said Rodolphe Gelin, who works alongside Dr. Luc JULIA , Scientific Director at Renault Group, to engage with a community of 400 developers, users, and facilitators, on such topics.

"It goes without saying that a tool like ChatGPT arouses people’s interest, curiosity and even some fears," added Sylvain Decaux , Business-to-Employee Manager at Renault Group IT. "The important thing is to frame the conversation because, as with any technological shift, it is essential to go with the change and help think about it critically." AI is becoming more widespread and commonplace at a quickening pace, which goes hand in hand with a growing number of issues, particularly in terms of reliability, regulation, data security, and intellectual property. Renault Group is strongly involved in defining the regulatory framework, as seen through its work as founding member of Confiance.ai, a French collective of people from industry and academia working rolling out an AI that is trustworthy, legible, and even certifiable.

"As with any new tool, it requires careful experimentation," said Sylvain Decaux. "There is real enthusiasm and interest, as proven with each new use case. There need to be a conversation and increased awareness to come up with a secure framework in which those experiments can be conducted. We have — and have had for some time — all the in-house skills to tackle these complex subjects internally. But AI is rapidly maturing, and it is a sound idea to approach its potential deployment in our own way, in a way that is in line with Renault Group's principles and values. Whatever the case, it will always serve to enhance both efficiency and safety.”

To this extent, the end of the year will see the launch of a knowledge-management project involving a search-engine overhaul and two forms of AI: semantic indexing AI that can analyse textual content of document databases and understand their meaning, to detect relevant information beyond simple keyword matches; and conversational AI to iteratively refine searches through natural-language conversation. "We are also keeping a close eye on what software vendors have on offer to add 'copilot' features to their tools. For example, with generative AI that can come up with draft texts, documents, images and computer code, or even act as a real personal assistant that will help organize the working day or summarize messages received during your absence."

“It's a bottom-up approach,” explained Rodolphe Gelin. “We start by working with people on the ground, learning about their needs before exploring a suitable technical solution.”

This phase also serves an educational objective, as Anthony Vouillon specified: “AI offers many reskilling opportunities, especially for people working in jobs on the slow decline. We have our work cut out for us to set up training courses that easily adapt where each person is in their life and career. But our aim is to teach AI as a skill for all, not to keep it within the hands of a select few.”

Instead of driving a revolution, AI is accelerating change. Without a doubt, it is yet another for a next-generation automotive company. A tool, but not a purpose. It must therefore be assessed on a case-bycase basis and deployed where it brings real added value.

Follow us to keep up to date with our news!