Artificial intelligence is inviting itself on board: with autonomous cars, image recognition and real-time data processing, the technological revolution is underway in the automotive industry. But what real opportunities does it open up? And what limits still need to be overcome?

At the Nexus 2025 tech confab on 17 June--organised by The Dots and Maison Moderne, with the support of the government, the City of Luxembourg, the Chamber of Commerce and Luxinnovation--four experts shared their views on this future already under construction.

Around the table: Ewald Liess, co-CEO and CRO of Teraki GmbH; Serge Lambermont, CEO and founder of Resembler.AI; Robert Sykora, director EMEA of Ohmio; and Valerie Zapico, CEO of Valkuren.

“A good autonomous car must continue to function even without a connection”

First observation: AI has its place in the vehicle... provided it does not become its master. For Liess, the challenge is clear. “A good autonomous car must continue to function even without a connection. The cloud improves the experience, but reliability depends on onboard intelligence.”

This approach meets the need for reliability in situations where the connection is not always stable. But it doesn’t solve everything: the machine must also understand its environment.

“Say a child runs along the road, followed by its mother. A human driver immediately understands what is happening. Can an AI do the same?” asks Lambermont. Whilst Level 4 autonomous cars are already more reliable than many human drivers, their ability to anticipate the unpredictable remains limited.

Onboard dialogue

This is where the human factor resurfaces, even in an autonomous system. Zapico is already imagining an in-car voice assistant capable of interacting with both the driver and the external environment: pedestrians, police officers, etc.

This human-machine interaction is also a question of trust. “Today, people are not ready to completely let go of the steering wheel. You need a human on board to reassure them and answer any questions,” explains Sykora. “Acceptance is the key to everything.”

It’s a transition that will require education and transparency. “Humans don’t like change,” admits Zapico. “We have to explain to them what’s going on in the black box.”

But how far should we let the machine decide? “We’ve all seen an AI system ‘hallucinate.’ If this happens whilst driving, the consequences can be catastrophic,” warns Liess. “Regulation must guarantee 100% safety. AI is difficult to supervise and implement, but without it, we won’t make any progress.”

There remains the big ethical question raised by Lambermont: even if AI proves to be a thousand times more reliable than humans, what should be done when it makes a mistake that a human could have avoided? The question remains unanswered, but it will quickly become obvious as the technology progresses.

A ten-year horizon

For large-scale deployment, we will still have to wait. “In controlled areas such as airports or dedicated bus lanes, we can imagine adoption within three to five years,” says Liess. “But for a taxi picking me up and dropping me off at the airport, it will be at least ten years.” It’s a time horizon shared by Sykora.

Zapico, for her part, is betting on a hybrid model, where autonomous functions will cohabit with human driving in well-supervised areas, such as in the city.

But the transformation goes beyond the vehicle. “AI will revolutionise our businesses over the next three to five years,” concludes Lambermont. “Concrete applications are multiplying, including outside mobility. This is just the beginning.”

This article was originally published in French.