AI Is Changing Car Design From The Inside Out

While car design is inherently a creative field, it is also fundamentally rooted in applied knowledge—an area where Artificial Intelligence excels.

However, the conversation shouldn't be about the superficial aspects of visual design. It's now trivially easy to prompt a generative AI with 'photo rendering of 2028 Porsche 911' and get a stunning result. Unfortunately, this ease is leading some cost-cutting executives to sideline skilled human designers, which is a significant loss for creativity.

Beyond Visuals The Engineering Revolution

The truly transformative power of AI in the automotive world lies in engineering design. An early and fascinating example of this is bio-mimicry. Nature, through millions of years of evolution, has perfected the design of animal bones to be incredibly strong and lightweight for their specific functions, despite looking nothing like conventional geometric structures. Mercedes has embraced this concept by training AI software on zoological data to develop advanced, optimized suspension arms.

When AI Writes the Code That Drives You

AI's influence now extends into the realm of software development. The automaker Xpeng claims its G6 model operates using AI. According to company founder He Xiaopeng, their proprietary AI, named XBrain, is "fed and trained" with vast amounts of information to develop its own nuanced understanding of traffic.

Remarkably, this AI software is itself coded by another AI. This creates a fascinating chain: human engineers at Xpeng have coded an AI (XBrain) that in turn codes the software that assists the human driver. With three layers of software separating the engineer from the end-user, Xpeng is positioning itself as a leader in automotive software. This prowess is so evident that the VW Group, which has struggled with its software in the Chinese market, is now adopting Xpeng's code as the backbone for its future cars in the region.

A Reality Check on the Road

But what does AI-assisted driving mean in practice? Essentially, it means self-learning, both at the individual vehicle level and collectively in the cloud. In theory, this allows a car to improve its performance over time.

This theory was put to the test on the roads of Kent in the UK. A British-spec Xpeng G6, ready for sale, exhibited a terribly clunky driver-assist suite. The system made sudden, late, and twitchy decisions, struggling to steer a smooth path. It's clear that while it may perform well on Chinese roads, it still has a great deal of self-learning to do in other environments. The real test will be to drive it again in a few months to see if its AI has genuinely learned and improved. This presents a new challenge for road testers but could ultimately benefit car buyers if the technology lives up to its promise.